Raise error for misconfigured 3D physics diagnostics

[spencerkclark]

This is now the output upon model termination if this occurs. As in #132, the previous termination would be an uninformative segmentation fault.

FATAL from PE     0: FV3GFS_io::fv3gfs_diag_register Outputting 3D diagnostics from the physics requires gfs_physics_nml.ldiag3d be set to true.


FATAL from PE     1: FV3GFS_io::fv3gfs_diag_register Outputting 3D diagnostics from the physics requires gfs_physics_nml.ldiag3d be set to true.


FATAL from PE     0: FV3GFS_io::fv3gfs_diag_register Outputting 3D diagnostics from the physics requires gfs_physics_nml.ldiag3d be set to true.

application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 1

===================================================================================
=   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
=   PID 13 RUNNING AT 8fdbee4d008c
=   EXIT CODE: 1
=   CLEANING UP REMAINING PROCESSES
=   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
===================================================================================

[nbren12]

Related to #130

[nbren12]

LGTM. Thanks. Could you shorten the title? I'd also be okay with just setting ldiag3d=.true. and not killing the model.

[spencerkclark]

Could you shorten the title?

Thanks, sure, that will produce a cleaner commit message.

I'd also be okay with just setting ldiag3d=.true. and not killing the model.

It's not this simple unfortunately, because currently the decision to allocate memory for the 3D diagnostics buckets is made before the diagnostics themselves are registered. So if we changed ldiag3d to .true. after we registered the diagnostics (read: parsed the diag_table), it would be too late. We could probably refactor things so that it didn't have to be this way, but I'm not sure if it's worth it.

[oelbert]

LGTM