fix(fluids): Remove comm arg from CeedOperatorCreateLocalVecs

[jrwrigh]

Vecs created this way should always be purely local and thus always use PETSC_COMM_SELF.

This caused issues when using the spanwise statistics in parallel:

[0]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------
[0]PETSC ERROR: Invalid argument
[0]PETSC ERROR: Int value must be same on all processes, argument # 3
[0]PETSC ERROR: #1 VecSetSizes() at /lus/gecko/projects/PHASTA_aesp_CNDA/petsc-kjansen-fork1015002/src/vec/vec/interface/vector.c:1467
[0]PETSC ERROR: #2 CeedOperatorCreateLocalVecs() at /lus/gecko/projects/PHASTA_aesp_CNDA/libCEED_1015002/examples/fluids/src/petsc_ops.c:326
[0]PETSC ERROR: #3 GetQuadratureCoords() at /lus/gecko/projects/PHASTA_aesp_CNDA/libCEED_1015002/examples/fluids/src/turb_spanstats.c:169
[0]PETSC ERROR: #4 CreateStatsSF() at /lus/gecko/projects/PHASTA_aesp_CNDA/libCEED_1015002/examples/fluids/src/turb_spanstats.c:262
[0]PETSC ERROR: #5 TurbulenceStatisticsSetup() at /lus/gecko/projects/PHASTA_aesp_CNDA/libCEED_1015002/examples/fluids/src/turb_spanstats.c:494
[0]PETSC ERROR: #6 SetupLibceed() at /lus/gecko/projects/PHASTA_aesp_CNDA/libCEED_1015002/examples/fluids/src/setuplibceed.c:413
[0]PETSC ERROR: #7 main() at /lus/gecko/projects/PHASTA_aesp_CNDA/libCEED_1015002/examples/fluids/navierstokes.c:199

[jedbrown]

Communicators can have attributes, so there can be reason to use a communicator of size 1 that is not literally PETSC_COMM_SELF. VecCreateSeq still has a communicator argument, for example.

[jrwrigh]

Ok. Do you think we should keep the comm argument then?

[jedbrown]

That would be my preference, and check that the communicator size is 1.

[jrwrigh]

Superseded by #1570