- aligns with the resulting VTK subdirectory name

ENH: adjust tokenizing

See merge request !452

- profit from some of the recent modifications to parser expansion

TUT: adjust some #eval statements for less clutter

- reworked from the openfoam.org wordAndDictionary version.

Allows, for example, named entries in topoSet.

* removed internal upper limit on word/string length for parsed input.

  - Although it has not caused many problems, no reason to retain
    these limits.
  - simplify some of the internal logic for reading string-like items.
  - localize parsers for better separation from the header

  - expose new function seekCommentEnd_Cstyle(), as useful
    handler of C-style comments

* exclude imbalanced closing ')' from word/variable

  - previously included this into the word/variable, but makes more
    sense to leave on the parser for the following token.

    Prevents content like 'vector (10 20 $zmax);' from being parsed
    as '$zmax)' instead of as '$zmax' followed by a ')'.
    No conceivable reason that the former would actually be desirable,
    but can still be obtained with brace notation: Eg, '${zmax)}'

* consistent handling of ${{ ... }} expressions

  - within a dictionary content, the following construct was
    incorrectly processed:

        value ${{2*sqrt(0.5)}};

    Complains about no dictionary/env variable "{2*sqrt(0.5)}"

    Now trap expressions directly and assign their own token type
    while reading. Later expansion can then be properly passed to
    the exprDriver (evalEntry) instead of incorrectly trying
    variable expansion.

    Does not alter the use of expressions embedded within other
    expansions. Eg, "file${{10*2}}"

* improve #eval { ... } brace slurping

  - the initial implementation of this was rudimentary and simply
    grabbed everything until the next '}'.  Now continue to grab
    content until braces are properly balanced

    Eg, the content:   value #eval{${radius}*2};

    would have previously terminated prematurely with "${radius" for
    the expression!

NOTE:
    both the ${{ expr }} parsed input and the #eval { ... } input
    discard C/C++ comments during reading to reduce intermediate
    overhead for content that will be discarded before evaluation
    anyhow.

* tighten recognition of verbatim strings and expressions.

  - parser was previously sloppy and would have accepted content such
    as "# { ..." (for example) as an verbatim string introducer.
    Now only accept parse if there are no intermediate characters
    discarded.

- minor simplification of #if/#endif handling

ENH: improve input robustness with negative-prefixed expansions (#2095)

- especially in blockMeshDict it is useful to negate an value directly.
  Eg,
  ```
     xmax  100;
     xmin  -$xmax;
  ```
  However, this fails since the dictionary expansion is a two-step
  process of tokenization followed by expansion. After the expansion
  the given input would now be the following:
  ```
     xmax  100;
     xmin  - 100;
  ```
  and retrieving a scalar value for 'xmin' fails.

  Counteract this by being more generous on tokenized input when
  attempting to retrieve a label or scalar value.
  If a '-' is found where a number is expected, use it to negate the
  subsequent value.

  The previous solution was to invoke an 'eval':
  ```
     xmax  100;
     xmin  #eval{-$xmax};
  ```
  which adds additional clutter.

ENH: splitMeshRegions: combine cellZones. See #2046

See merge request !448

refactor and extend handling of faSchemes/fvSchemes

See merge request !449

- use common base for handling scheme lookups.
  Extended to support local injection of schemes into the dictionaries
  during runtime.

- add const qualifiers on C() methods

collected changes for Lists, faces and PrimitivePatch

See merge request !446

- intended for the following type of use:

      auto oldHandler = fileHandler(fileOperation::NewUncollated());

      ... do something that only works with uncollated

      // Restore previous (if any)
      if (oldHandler)
      {
          fileHandler(std::move(oldHandler));
      }

ENH: make fileOperation distributed(bool) mutable

- use is "static-like" and akin to Pstream::parRun(bool),
  thus allow toggling of the switch without a const_cast

- use polyMesh::defaultRegion instead of fvMesh::defaultRegion and
  others via static inheritance. (92 vs ~25 occurrences)

- return internalEdges() and boundaryEdges() sub lists directly

- calculate and return boundaryFaces() to identify faces attached to
  boundary edges.

- minor code cleanup, and add PrimitivePatchBase class for
  non-templated code.

STYLE: mark unused parameter in globalMeshData mergePoints

- copy assignment from indirect list to SubList and Field

- this constructor was added for similarity with std::vector,
  but continues to cause various annoyances.

  The main problem is that the templated parameter tends to grab
  anything that is not a perfect match for other constructors.

  Typically seen with two integers (in 64-bit mode), but various other
  cases as well.

  If required, the ListOps::create() function provides a lengthier
  alternative but one that can also incorporate transformations.
  Eg,

      pointField pts = ....;

      List<scalar> mags
      (
          List<scalar>::create
          (
              pts.begin(),
              pts.end(),
              [](const vector& v){ return magSqr(v); }
      );

- improves interchangeability of List vs IndirectList

- additional rcEdge(), rcEdges() methods for reverse order walk

- accept generic edge() method as alternative to faceEdge() for
  single edge retrieval.

- edge() method with points -> returns the vector

- reduce the number of operations in edgeDirection methods

DEFEATURE: remove longestEdge global function

- deprecated and replaced by face::longestEdge() method (2017-04)

- gcc-4.8.5 does not respect the default parameter,
  thus specify format explicitly

Historically the "geometry" for static meshes was placed under
directly in the EnSight case directory.

Eg,

     ensight/ensight.case
     ensight/geometry
     ensight/data/000001
     ensight/data/000002 ...

This generally works ok, but relocating it to a constant data directory
Eg
     ensight/ensight.case
     ensight/data/constant/geometry
     ensight/data/000001
     ensight/data/000002 ...

Improves handling and avoids potential collisions when adding in
additional mesh regions

- reduces code, simplifies creation of new, specialized polyData
  writers.

- new templated vtk::GenericPatchWriter, which adds support for
  writing both uindirectPrimitivePatch + indirectPrimitivePatch types.

- handle geometric fields separately from regular fields with
    * vtk:GenericPatchGeoFieldsWriter
    * vtk:indirectPatchGeoFieldsWriter
    * vtk:uindirectPatchGeoFieldsWriter

- can be used for block-like meshes that are not aligned with the global
  coordinate directions. Alternatively, for general testing purposes.

  Example,

    method  simple;
    coeffs
    {
        n       ( 2 2 2 );
        transform
        {
            origin  (-0.15 0.15 0);
            e1      (1 1 0);
            e3      (0 0 1);
        }
    }

add multi-region handling for checkMesh etc,  centralized the region handling #2072

See merge request !445

Step 1.
    include "addAllRegionOptions.H"

    Adds the -allRegions, -regions and -region options to argList.

Step 2.
    include "getAllRegionOptions.H"

    Processes the options with -allRegions selecting everything
    from the regionProperties.

    OR use -regions to specify multiple regions (from
       regionProperties), and can also contain regular expressions

    OR use the -region option

    Specifying a single -regions NAME (not a regular expresssion)
    is the same as -region NAME and doesn't use regionProperties

    Creates a `wordList regionNames`

Step 3.
    Do something with the region names.
    Either directly, or quite commonly with the following

    include "createNamedMeshes.H"

    Creates a `PtrList<fvMesh> meshes`

STYLE: add description to some central include files

- useful for establishing and preallocating a max buffer size
  when reading from sub-procs

improve flexiblity of IOdictionary construction, use of ITstream, Function1

See merge request !444

- make regionName an optional constructor parameter, which eliminates
  a separate set of constructors and construction tables. Adjust
  internals to treat a missing/empty regionName as a no-op.

- pass in fallback dictionary content via new IOdictionary constructor
  with a pointer

ENH: further relax check for matching number of processor dirs

- if the "numberOfSubdomains" entry is missing (or even zero)
  ignore checks of processor dirs as meaningless.