Set default configuration to not setup Pyre
Added note in examples directory regarding Pyre and Cookbooks

can be assigned without setting ibuoy=1 and the buoyancy number to zero. 

The bug is harmless though.

I checked the code history. This bug exists since CitcomS 1.0, but not in
CitcomCU nor CitcomT.

fixed a bug that h5close_field() didn't have a return statement, reported by Nicolas Flament in issue611.

fixed a bug that when the perturblayer is on the internal boundary, the temperature perturbation wasn't added to the lower processor.

changes on my side.

- A new input parameter 'vtk_format', which can be either 'ascii' (default) or
  'binary'. When 'binary', gzip and base64 encoding are used.
- The node ordering is changed to CitcomS native ordering. The consequence is the
  axis are rotated as z in CitcomS is mapped to X in vtk, and x to Y, y to Z.
- More data can be outputted in vtk format
- Using .pvts file for regional model; .vtm and .visit file for global model
  (.vtm format provided by Tobias Hoeink at Rice, .visit format provided by Kat
   at visitusers.org/forum)

enforce an iterative solution to the velocity fields even if linear
rheologies were chosen. This is useful for debugging/robustness tests.

Those are didentical

No commit message

screwed up tracer-based runs with  large number of processors (12x3x3x2). 
Weird, but oh well. Someone advise why this didn't work?

to compare model computations with previous versions. 

assignment apparently screwed up.

Fixed a few compiler complaints. 

Made sure that atemp is always set in Viscosity_structures, making
mat_control > 0 feasible, not just 0 or 1

has a check for 

ROW==E->lmesh.NOZ[level] && E->parallel.me_loc[3]==E->parallel.nprocz-1

while the full version has 


ROW==E->mesh.NOZ[level]) && (E->parallel.me_loc[3]==E->parallel.nprocz-1) )