- patch from Bernhard Gschaider

- consistent use of watchIndex vs watchFd

Feature read now

See merge request !102

Fixes #420.

The typical topology is the one where boundary faces share non-consecutive
points (checkMesh reports this as 'Number of faces with non-consecutive shared points')
This is handled by no-extruding any of the vertices of both faces. Fixes #391.

- use similar semantics as used for dealing with metis, scotch

- This can be used as a convenient alternative to comparing against end().
  Eg,

    dictionaryConstructorTable::iterator cstrIter =
        dictionaryConstructorTablePtr_->find(methodType);

    if (cstrIter.found())
    {
        ...
    }
  vs.
    if (cstrIter != dictionaryConstructorTablePtr_->end())
    {
        ...
    }

- reduce the amount of checking.
  Equivalent logic to what HashTable has.

- remove stray canonicalSize declaration

- 1st problem arises when there are edges, but edgeNormals is empty.
  The UIndirectList fails (zero elements, non-zero addressing)

- further problem occurs if there is a mismatch in the number of edges
  and edges normals (incorrect indexing on loop).

- can be useful with compiling additional OpenFOAM programs
  that use FOAM_USER_APPBIN, FOAM_USER_LIBBIN for their build,
  to avoid conflicts with the normal user bin/lib files.

- or to force relocation of FOAM_SITE_APPBIN, FOAM_SITE_LIBBIN
  during packaging of OpenFOAM

- consistent with use of FOAM_SOLVERS, and reduces reliance on the FOAM_APP
  env variable

- had occasional remnant use of FOAM_SRC

Fixed by setting flag which then gets queried by Time. Fixes #420.

- useful for builds that don't use the ThirdParty directory at all

- they don't use scotch/metis anyhow, which are the only things
  placed in FOAM_EXT_LIBBIN

- just check WM_PROJECT_DIR instead.

- provide a fallback value when FOAM_EXT_LIBBIN might actually be needed.

Only strictly need FOAM_EXT_LIBBIN for scotch/metis decomposition, and
when these are actually supplied by ThirdParty.
All other ThirdParty dependencies are referenced by BOOST_ARCH_PATH etc.

Can therefore drop the FOAM_EXT_LIBBIN dependency for VTK-related
things, which do not use scotch/metis anyhow.

- has different locations on different systems
  (eg, /usr/bin/lsof or /sbin/lsof)

- this implies that jobControl is a user-resource for OpenFOAM.
  It was previously located under $WM_PROJECT_INST_DIR/jobControl,
  but few users will have write access there.

- an unset FOAM_JOB_DIR variable is treated as "~/.OpenFOAM/jobControl",
  which can partially reduce environment clutter.

- provide argList::noJobInfo() to conveniently suppress job-info on an
  individual basis for short-running utilities (eg, foamListTimes) to
  avoid unneeded clutter.

- previously used the value of $WM_PROJECT, but this is potentially
  inconsistent with what Foam::etcFiles uses.

- ensure proper and sensible handling of empty names.
  Eg, isDir(""), isFile("") are no-ops, and avoid file-stat

- rmDir:
  * optional 'silent' option to suppress messages.
  * removes all possible sub-entries, instead of just giving up on
    the first problem encountered.

- reduced code duplication in etcFiles

ENH: provide WM_USER_RESOURCE_DIRNAME define (in foamVersion.H)

- this is still a hard-coded value, but at least centrally available

- minor bug: 'wclean empty' may have had issues with logical vs physical path

- permit SYSTEMMPI user adjustments via
      etc/config.{csh,sh}/mpi-system
  This can be a convenient place for setting up SYSTEMMPI for OpenFOAM
  without adjusting bashrc, prefs.sh ...

- add a USERMPI type. This represents any generic mpi implementation.
  The user is responsible for supplying an appropriate
  wmake/rules/General/mplibUSERMPI file and managing all settings.

  This type of setup can be useful in combination with specific build
  systems (SPACK, EASYBUILD, etc) or module systems for which the MPI
  variant is part of the installed configuration.