- remove epsilonSmall, omegaSmall
- k0/epsilon0/omega0 become kMin/epsilonMin/omegaMin
- add qMin/zetaMin for consistency

These files still need some attention:
    dynOneEqEddy.C
    NonlinearKEShih.C
    settlingFoam

BUG: incompressible::LESModels:dynOneEqEddy::correct()
- avoid tmp field destruction for consistency with the compressible
  version

Possible TODO:
   - set kMin to zero (instead of SMALL) and introduce kSmall
     to avoid division by zero

ENH: Fixed documentation comments for the k0, epsilon0, omega0

BUG: omegaSmall was not being used in compressible RAS models

Otherwise all $PATH will pick up the local one which will not be synchronised with $path

Reverted to using additional variable.

Conflicts:
	etc/settings.csh
	etc/settings.sh
	src/parallel/decompose/Allwmake

ptscotch - compiles into ptscotchDecomp. All thirdparty decompositionMethods
now moved out of decompositionMethods so add them explicitly to link line
for programs that need them (decomposePar, snappyHexMesh etc.)

COMP: drop gcc-4.2 settings

Added compressibility support to MRFZone.
Required generalisation of the oneField concept.
Also clean-up the use of one and zero for consistency.
Simple mixerVessel2D tutorial case supplied.

- the path is accesible either way, but use setenv for consistency with
  other environment variables. It also simplifies foamCleanPath again.

For example, within the user ~/.OpenFOAM/<VER>/prefs.sh:

    foamPrefs=`$WM_PROJECT_DIR/bin/foamEtcFile -m go prefs.sh` \
        && _foamSource $foamPrefs

could be used to write a personal adjustment that also includes the
site-wide prefs.

BUG: compilerInstall was getting overwritten in settings.{csh,sh}