3 - SPM/Dist_basic_routines

Make sure some of the basic routines are working with distributed matrices, or prevent issue by returning an error: 
 * SpmPrint
 * Spm2Dense
 * SpmConvert
 * SpmGenFakeValues

Some of these routines have also been largely factorized to try to simplify them.

See merge request !23

Tony Delarue authored 4 years ago and Mathieu Faverge committed 4 years ago

   - Generation of fake values
   - Format conversion (More cases are supported in shared memory too)
   - Factorization of the conversion to dense format to support any dof (Not available in distributed)
   - Factorization of the print function to support any dof (Still require to provide different output file descriptors in distributed)

Hotfix/spm scatter

See merge request !30

2 - SPM/Scatter

See merge request !22

Fix issues with the actual spm in order to make it works in distributed

 This PR fixes minor issues that were not impacting the library in the shared memory case, but that would have a large impact on the distributed case.

 * Add a spmInitDist() function to initialize an spm with a specific communicator (MPI_COMM_WORLD by default)
 * Add a function to compute the glob2loc array when needed.

See merge request !29

Fix warnings and issues raised by coverity

See merge request !20

Upgrade CI scripts and Ctests to prepare the MPI version

See merge request !19

Update cmake

See merge request !17

Hotfix/ijv merge

See merge request !16

Fix Conjuguate Transpose on hermitian matrices

See merge request !15

Rules precision

Update the rule precision scripts to get the latest one with local dictionaries.

See merge request !14