- Less looping when detecting lagrangian clouds and their fields.

- Avoid using Time::setTime() and IOobjectList in tight loops.
  They both kill performance immensely.

ENH: provide a -noLagrangian option to foamToEnsight and foamToEnsightParts
for even more control.

- More informative than a horizonal line, can help when debugging.

STYLE: remove unused write field methods from ensightParts

- The new field needs initialization with a dimensioned<Type> not just
  the dimensionSet.

- The new field was also incorrectly being registered, which could
  cause issues later.

- affects foamToEnsightParts, sampled surfaces

- Use ensightPTraits mechanism throughout to avoid this issue

Old code:
    Found 10990 time steps
    Search for moving mesh ... no moving mesh detected.
    Startup in 329.09 s

Updated:
    Found 10990 time steps
    Search for moving mesh ... no moving mesh detected.
    Startup in 1.6 s

- Cause was checking "polyMesh/points" via an IOobject.
  Short-circuit with a check for a polyMesh/ directory first.

  Limit the check to the master-node as well to further reduce
  load on the file-system.

------------------------------

ENH: improve per-step conversion times for foamToEnsight.

Old code:
    Converting 11001 time steps
    Time [0] = 0       Wrote in 1.53 s
    Time [1] = 1       Wrote in 1.52 s
    ...
    Time [100] = 100   Elapsed time 205.35 s

Updated:
    Converting 11001 time steps
    Time [0] = 0       Wrote in 1.4 s
    Time [1] = 1       Wrote in 0.07 s
    ...
    Time [100] = 100   Elapsed time 42.4 s

- Speedup by hashing test results from the first conversion step
  instead of checking each time.

  Check data on all nodes to avoid problems with incomplete writes.

------------------------------

BUG: moving mesh detection failed for foamToEnsightParts

- adjusted to agree with updated foamToEnsight

------------------------------

Note:

- foamToEnsightParts (serial) still has about twice the throughput of
  foamToEnsight.

BUG: output field name instead of field (fixes #224)



See merge request !59

- bugfix (empty patches), and added detection of steady-state
  scheme.

Caveat: when called via execFlowFunctionObjects will always produce a
  zero field, since the oldTime field is not available for this mode.

Extrapolate internal field to walls for post-processing.
Uses as new syntax for handling the naming of multiple fields.

The input fields are selected via a wordReList.
For example,

    fields      (U "(T|k|epsilon|omega)");

The names of the resulting output fields use placeholder tokens for
flexibility. For example,

    result      zeroGradient(@@);

The '@@' placeholder is replaced by the name of the input field.
Eg,
    fields      (U T);
    result      zeroGradient(@@);
->  zeroGradient(U), zeroGradient(T)

Or,
    fields      (U T);
    result      @@nearWall;
->  UnearWall, TnearWall

NOTE:
    The function object will skip over fields that only have
    processor, empty, zeroGradient patches. The operation does not
    make much sense for these, and it avoids inadvertently
    re-processing fields twice.

- implemented using magSqr() instead of sqr().
  For scalar fields they are the same, but can be useful
  if this function object is extended for more field types.

- Default is a width of 8 characters, but this can be extended up to 31
  characters via the '-width' command-line option.

- Now use a similar structure as foamToEnsightParts for the masking.
  This reduces the clutter within the directory, makes it easier to
  selectively delete some time steps (using shell commands).

- Added in a "time" information data in each sub-directory to
  make it possible to reconstruct the case file with an external
  script.

- Conversion of cloud data should now also work in parallel
  (may need more testing).

- Support binary output for cloud data.

- Better avoidance of illegal ensight variable names.
  But still partially incomplete (due to patch fields).

==================================================
Example of NEW file structure:

    EnSight/verticalChannel.case        # case name
    EnSight/geometry                    # for non-moving geometry

    EnSight/data/                       # time-varying data
    EnSight/data/00000000/
    EnSight/data/00000001/
    ...

  Fields are stored by name within the data/********/ directories:

    EnSight/data/00000001/time          # human-readable time info
    EnSight/data/00000001/U
    EnSight/data/00000001/p
    ...
    EnSight/data/00000001/geometry      # for moving geometry

  Clouds are stored at the next sub-directory level:

    EnSight/data/00000001/lagrangian/<cloudName>/positions
    EnSight/data/00000001/lagrangian/<cloudName>/U
    ...

==================================================
The old structure was significantly more cluttered:

    EnSight/verticalChannel.case
    EnSight/verticalChannel.0000.mesh
    EnSight/verticalChannel.0001.p
    EnSight/verticalChannel.0001.<cloudName>
    EnSight/verticalChannel.0001.<cloudName>.U

==================================================

- 64-bit builds of gcc/mpfr/gmp use lib64/ for their installation path.
  Use this for the wmake rules as well.

- spurious use of floatScalar instead of scalar for processor weights.

STYLE: partially updated dummy metis.h to reflect metis-5.1.0 API

Fixes #66.

- reported by Bruno

- Also reported as http://bugs.openfoam.org/view.php?id=2188

- add corrsponding testing into surfaceMeshConvertTesting too

BUG: ensightSurfaceReader - enabled reading of data from sub-directory and updated field mask (See #215)

- gearing to make it more reusable in triSurface

This was going wrong in a fair few cases where you would get regions which
were delimited by a combination of intersections with unnamed surfaces (wall)
and named surfaces (faceZone surface) but not with either of these sets
individually. This would cause there to be unvisited cells (zone = -2)
next to a visited cells but separated by a named surface only (and not an
unnamed surface). Fixed by using direct surface intersection in the removal
part. Related to #66.