* Fix #474
* Add cgns2nek
* Add blasLapack as 3rd party
* Add nekconfig

* direct solve for p0th
* fix token parser to get restart file in par
* add ifvvisp logical

* fix komg evalutation logic
* move neknek_setup after geom_reset + enforce fixgeom

* replaced depcreated mpi_attr_get
* disable CHT and enable scalars for neknek
* print min/max GLL spacings

* Add semg_amg_hypre preconditioner
* Fix nullspace handling for fem_amg_hypre
* Add Cray and ARM compiler support 
* Set default optimization level for 3rd-party
* Use dynamic memory allocation for exo2nek and improved sideset handling
* Add gencon
* Update to latest HYPRE version
* Change interp interface to support multisession
* Add parRSB
* Add distributed proc mapping
* Initialize ifdgfld #511
* Update tools to core/byte.c (#547)
* Remote SIZE files of all short_tests
* Remove GTP
* Add max iteration count for all fields (#534)

* add multi-seesion math routines ms_glmax etc.
* getting rid of userchk_set_xfer
* add a routine to interpolate specific variable between sessions
* add support for passive scalars 
* always use valint in userbc
* call mutlimesh_create and set defaults outside .usr

    * eliminate PPLIST symbol NEKNEK (instead read number of sessions from SESSION.NAME)
    * single routine to create communicators for all supported modes
    * add global tensor product plane and multisession gather scatter operations

* Replace KovSteady by KovStokes
* Changed default of short tests to 2 rank
* Add FsHydro short test
* Adjust to a meaningful tolerance for velocity
* Add projection for Helm3
* Add incl_def as short test
* Skip reading igroup
* Fix reading errors in nekmerge
* Add axhmsf to axhm timer
* Use env-vars instead of parsing maketools/makenek
* Correct max number of blocks for exo2nek
* Test against analytical solution in axi short test
* Set lastep even when IO is disabled
* Remove support for Cray compilers
* Remove some timers as the overhead is too large
* Add runtime check for eddy
* Fix to compile tools with IBM compilers
* Do a clean before we start the build process for short tests
* Create .f userfile with make instead of makenek

* switch for FC instead of F77
* delete k-eps model
* build CVODE on demand
* fix gfldr for PnPn2
* disable checkpoints for iostep=0 or iotime=0
* make par reader more robust
* add support for elasticity solver in par file
* correct number of RK4 substeps for varying dt
* switch to new par v1.1
* introduce user specified Jacobian-vector product for CVODE
* add residual tolerance for field
* allow user to specify maximum number of elements in maketools
* fix byte_swapping issues introduced by #378
* add new projection scheme
* change nx1->lx1 etc.
* use projection in conj-ht

* compile amg_hype and exo2nek including external libs
* refactor PAPI interface to count FLOPs
* changed .par keyword AMG to SEMG_AMG
* set param(41-44)=0 by default irrespective of .rea
* compile tools with BIGMEM by default
* add counters users, users, addcol3, col2, col3, add2s2
* remove pressure solve from helm-counter
* enable mxm-counter
* fix restart issue with residual projection (#361)
* remove avg from tools
* add .his to cpn, mvn

This reverts commit 7f2440ce.

Read .re2 in parallel by default
Make MPIIO a runtime option instead of a preprocessor symbol

- Delete MOAB support
- Delete compile test as the executables may work on the compute node only
- Move stdout to the corresponding .re2 routines
- Make deformed elements the default
- Move makenek to bin to use it as default
- Exit makenek if any subcommands return a non-zero status
- Read his.in instead of hpts.in
- Output <casename>.his instead of hpts.out
- Make sure we always loop over nelt in fix_geom
- Delete old history point routines
- Do not call any compiled executables in makenek

- Add generic fld reader
- Print max memory usage on exit
- Remove internal usage of intpts wrapper
- Refactor qthermal handling