- end_value() corresponds to the infrequently used after() method, but
  with naming that corresponds better to iterator naming conventions.

  Eg,

     List<Type> list = ...;
     labelRange range = ...;

     std::transform
     (
         (list.data() + range.begin_value()),
         (list.data() + range.end_value()),

         outIter,
         op
     );

- promote min()/max() methods from labelRange to IntRange base class

STYLE: change timeSelector from "is-a" to "has-a" scalarRanges.

- resets min/max to be identical to the specified value,
  which can be more convenient (and slightly more efficient) than doing
  a full reset followed by add()

- additional MinMax intersects() query, which works like overlaps()
  but with exclusive checks at the ends

- provide MinMax::operator&=() to replace (unused) intersect() method

ENH: single/double value reset method for boundBox

- boundBox::operator&=() to replace (rarely used) intersect() method.
  Deprecate boundBox::intersect() to avoid confusion with various
  intersects() method

COMP: provide triangleFwd.H

- background: for some application it can be useful to have fully
  sorted points. i.e., sorted by x, followed by y, followed by z.

  The default VectorSpace 'operator<' compares *all*
  components. This is seen by the following comparisons

  1.  a = (-2.2 -3.3 -4.4)
      b = (-1.1 -2.2 3.3)

      (a < b) : True
      Each 'a' component is less than each 'b' component

  2.  a = (-2.2 -3.3 -4.4)
      b = (-2.2 3.3 4.4)

      (a < b) : False
      The a.x() is not less than b.x()

  The static definitions 'less_xyz', 'less_yzx', 'less_zxy'
  instead use comparison of the next components as tie breakers
  (like a lexicographic sort).
  - same type of definition that Pair and Tuple2 use.

      a = (-2.2 -3.3 -4.4)
      b = (-2.2 3.3 4.4)

      vector::less_xyz(a, b) : True
      The a.x() == b.x(), but a.y() < b.y()

   They can be used directly as comparators:

      pointField points = ...;

      std::sort(points.begin(), points.end(), vector::less_zxy...

- useful when a characteristic per-face search dimension is required.
  With PrimitivePatch we are certain to have consistent evaluations
  of the face centre.

STYLE: tag PrimitivePatch compatibility headers as such

STYLE: more consistent access for uniform list/fields

STYLE: combine templated/non-templated headers (reduced clutter)

STYLE: use hitPoint(const point&) combined setter

- same as setHit() + setPoint(const point&)

ENH: expose and use labelOctBits::pack method for addressing

- the old List_FOR_ALL macro only remained in use in relatively few
  places. Replace with the expanded equivalent and move the looping
  parameter out of the macro and give an explicit name (eg, loopLen)
  which simplifies the addition of any loop pragmas in the various
  TFOR_ALL... macros (for example).

ENH: simplify bookkeeping within MappedFile

- in places where direct reading from the std::stream is used,
  this method can be used to ensure that the OpenFOAM Sstream state
  is properly updated from the std::stream.

ENH: restrict stream renaming to ISstream

- non-const access was previously declared at the top-level (IOstream)
  but that not only added in potentially odd setting of the static
  fileName, but also meant that the OFstream name() could potentially
  be altered after opening a file and thus be inconsistent with the
  underlying file that had been opened.

  Now restrict name modification to ISstream (and ITstream
  counterpart). Does not affect any existing valid code.

STYLE: non-default OFstream destructor (for future file staging)

- remove old, unneeded -DUSE_OMP define.
- wmake -no-openmp option to add '~openmp' to WM_COMPILE_CONTROL

ENH: add bash completion handling for wmake

- defined locally, independent of wmake rules

- regression in develop, not other branches

Feature master coarsest multi masters

See merge request Development/openfoam!565

STYLE: simpler initialization

- construct boundBox from Pair<point> of min/max limits,
  make sortable

- additional bounding box intersections (linePointRef), add noexcept

- templated access for boundBox hex-corners
  (used to avoid temporary point field).
  Eg, unrolled plane/bound-box intersection with early exit

- bounding box grow() to expand box by absolute amounts
  Eg,

      bb.grow(ROOTVSMALL);   // Or: bb.grow(point::uniform(ROOTVSMALL));
  vs
      bb.min() -= point::uniform(ROOTVSMALL);
      bb.max() += point::uniform(ROOTVSMALL);

- treeBoundBox bounding box extend with two or three parameters.
  The three parameter version includes grow(...) for reduced writing.
  Eg,

      bb = bb.extend(rndGen, 1e-4, ROOTVSMALL);

  vs
      bb = bb.extend(rndGen, 1e-4);
      bb.min() -= point::uniform(ROOTVSMALL);
      bb.max() += point::uniform(ROOTVSMALL);

  This also permits use as const variables or parameter passing.
  Eg,

      const treeBoundBox bb
      (
          treeBoundBox(some_points).extend(rndGen, 1e-4, ROOTVSMALL)
      );

- box method on meshShapes (cell,edge,face,triangle,...)
  returns a Pair<point>.

  Can be used directly without dependency on boundBox,
  but the limits can also passed through to boundBox.

- Direct box calculation for cell, which walks the cell-faces and
  mesh-faces.  Direct calculation for face (#2609)

- with geometryOrder=1, calculate the edge normals from the adjacent
  faces (area-weighted, inverse distance squared) and also
  use that for the Le() calculation.

  Includes the contributions from processor edge neighbours, so it
  should be consistent on both sides.

  This new method (consider as 'beta') contrasts with the current
  standard method that first calculates area-weighted point normals
  and uses the average of them for the edge normal.

  Enable for testing either with a controlDict OptimisationSwitch entry
  "fa:geometryOrder", or on the command-line:

      solverName -opt-switch=fa:geometryOrder=1

- the Le vector is calculated from (edgeVec ^ edgeNorm)
  and should be oriented in direction (faceCentre -> edgeCentre).

  If, however, the edgeNorm value is bad for any reason, the
  cross-product falls apart and Le vector is calculated as a zero
  vector!

  For these cases, revert to using (faceCentre -> edgeCentre)
  as a better approximation than a zero vector.

  In the future, will very likely switch calculating the edge normals
  directly from the attached faces, instead of from the attached
  points as is currently done, which should improve robustness.

ENH: expose fa:geometryOrder as a registered OptimisationSwitch

ENN: reuse polyMesh data (eg, faceCentres) if possible in faMesh

STYLE: add code lambdas and static functions to isolate logic

ENH: extend rmDir to handle removal of empty directories only

- recursively remove directories that only contain other directories
  but no other contents. Treats dead links as non-content.

- stem(), replace_name(), replace_ext(), remove_ext() etc

- string::contains() method - similar to C++23 method

  Eg,
      if (keyword.contains('/')) ...
  vs
      if (keyword.find('/') != std::string::npos) ...

- similar to treeDataCell, doesn't need any meshTools components
- AABBTree under algorithms (like indexedOctree)

- construct based on db and mesh information from an existing field

- check movable() instead of isTmp() when reusing fields

STYLE: isolate check for reuse GeometricField into Detail namespace

- code remnant from separate lookup + construct of coordinateSystem
  (7b2bcfda).
  Apply consistent use of coordinateSystem::NewIfPresent to avoid
  these types of coding mishaps

- enables special purpose debugging.
  Its definition and behaviour are subject to change at any time.

- in continuation of #2565 (rotationCentre for surface output formats)
  it is helpful to also support READ_IF_PRESENT behaviour for the
  'origin' keyword.

  This can be safely used wherever the coordinate system definition
  is embedded within a sub-dictionary scope.

  Eg,
      dict1
      {
          coordinateSystem
          {
              origin (0 0 0);  // now optional here
              rotation ...;
          }
      }

   but remains mandatory if constructed without a sub-dict:

      dict2
      {
          origin (0 0 0);   // still mandatory
          e1  (1 0 0);
          e3  (0 0 1);
      }

   With this change, the "transform" sub-dictionary can written
   more naturally:

       formatOptions
       {
           vtk
           {
               scale 1000;  // m -> mm
               transform
               {
                   rotationCentre  (1 0 0);
                   rotation axisAngle;
                   axis    (0 0 1);
                   angle   -45;
               }
           }
       }

ENH: simplify handling of "coordinateSystem" dictionary lookups

- coordinateSystems::NewIfPresent method for optional entries:

    coordSysPtr_ = coordinateSystem::NewIfPresent(mesh, dict);

  Instead of

    if (dict.found(coordinateSystem::typeName, keyType::LITERAL))
    {
        coordSysPtr_ =
            coordinateSystem::New
            (
                mesh_,
                dict,
                coordinateSystem::typeName
            );
    }
    else
    {
        coordSysPtr_.reset();
    }

ENH: more consistent handling of priorities for binModels, forces (#2598)

- if the dictionaries are overspecified, give a 'coordinateSystem'
  entry a higher prioriy than the 'CofR' shortcuts.

  Was previously slightly inconsistent between the different models.