- the raw surface writer simply outputs x/y/z and field values.
  This additional flag allows recovery of some geometric information.

- optional user-specified output precision

Example,
```
formatOptions
{
    raw
    {
        normal      yes;
        precision   10;
    }
}
```

Kutalmış Berçin authored 4 years ago and Mark OLESEN committed 4 years ago

- centralises existing functions (erfInv, incGamma*, invIncGamma*).
  Provides a location for additional functions in the future.

- adjusted existing models to use these functions
  (e.g. distributionModels::normal)

- location and naming more consistent with other controls

    old: DebugSwitch  'level'
    new: InfoSwitch   'outputLevel'

- effectively 'steals' the pointer from the table but leaves its
  name as a placeholder

- eliminates a potentially invalid code branch.
  Since it essentially had the same internals as std::swap anyhow,
  make that more evident.

ENH: use std::swap for basic types

- makes it clearer that they do not rely on any special semantics

- since the wrapped cmake calls generally use the regular build
  locations, add in MPI information to properly handle changes
  in that as well. This makes it easier to build for multiple MPI
  instances.

- ensure surface writing is time-step and nFields aware.
  This avoids overwriting (ignoring) previous output fields.

- allow sampled surfaces to be used for weight fields as well.
  Not sure why this restriction was still there.

- remove old compatibility reading of orientedFields.
  Last used in v1612, now removed.

- only use face sampling. For surfaceFieldValue we can only do
  something meaningful with face values.

ENH: modify interface methods for surfaceWriter

- replace direct modification of values with setter methods.
  Eg,
     old: writer.isPointData() = true;
     new: writer.isPointData(true);

  This makes it possible to add internal hooks to catch state changes.

ENH: allow post-construction change to sampledSurface interpolation

- rename interpolate() method to isPointData() for consistency with
  other classes and to indicate that it is a query.

- additional isPointData(bool) setter method to change the expected
  representation type after construction

- remove 'interpolate' restriction on isoSurfacePoint which was
  previously flagged as an error but within sampledSurfaces can use
  sampleScheme cellPoint and obtain representative samples.
  Relax this restriction since this particular iso-surface algorithm
  is slated for removal in the foreseeable future.

1) Small modification to the tracking logic for detA zero.
2) Adding small vector displacement to locate function to avoid error where
particle is inserted at the cell centre.

- setup writer outside the data loop to ensure that the number of
  output fields is correct (VTK format).

- ignore 'interpolate' on sampled surfaces to ensure proper
  face sampling, never allow point sampling

BUG: incorrect debug-switch for sampledIsoSurface

PtrListOps and adjustments for sync

See merge request !422

- get: uses access operation to get values for each list item
  Example,
  PtrListOps::get(mesh.boundaryMesh(), nameOp<polyPatch>());

- names: the name() of each list item filtered for matches

- firstMatching: index of first item with a matching name()

- findMatching: indices of items with a matching match name()
  Example,
  PtrListOps::findMatching(mesh.boundaryMesh(), wordRes( ... ));

STYLE: deprecate transitional getNameOp, getTypeOp

- use nameOp, typeOp (word.H) instead

Was checking the old celli instead of the result of
re-finding the position. See also Foundation commit 50a965f8866683a81d79cbc7811af7333baf9d10.

If the 'writeFields' option is set in surfaceFieldValue, e.g.

    surface1
    {
        type        surfaceFieldValue;
        libs        (fieldFunctionObjects);
        operation   none;
        fields      (p);
        regionType  patch;
        name        walls;

        // Create a surface in VTK format
        writeFields yes;
        surfaceFormat vtk;
    }

... the surface can now be used in runTimePostProcessing, e.g.:

    surfaces
    {
        surfaceFieldValueOutput
        {
            type            functionObjectSurface;
            representation  surface;
            liveObject      no;
            field           p;
            colourBy        field;
            range           (0 120000);
            functionObject  surface1;
        }
    }

Note: setting 'liveObject' to 'no' to suppress warnings due to the surface
not being retrieved from the object registry (default = 'yes') - this surface
can [currently] only be read from disk.

- modification/continuation of 8d63073b and 5c1ec7ec (#595).

  Although this protected function is only used internally, the name
  `size(label)` is too easily confused with `resize(label)` and
  `setSize(label)`. The longer method name eliminates some ambiguity.
  Name consistent with PtrListDetail.

- leave size(label) method (for possible compatibility),
  but mark as deprecated

- improve sizing consistency for (Istream >> DynamicList)

STYLE: more consistent use of resize vs setSize in DynamicList

- more consistency between DynamicList and DynamicField.
  There were some inconsistencies in how construct with a size was
  interpreted.

STYLE: more consistent declaration/use of Swap

- add byteSize to FixedList and Matrix, for possible streaming

ENH: expose data/cdata for PackedList, bitSet

- a C-string is interpreted as a bool instead of as a word,
  so remove default parameter to eliminate ambiguity

- number of fields was not set.
- interpolated surfaces incorrectly written for all formats