Also remove a call to SELECT_JOBDATA_IONODES which is a no op.

With BG removed, this exposes that this function call had been reused
for select/alps for a different purpose. Rename the function to make
this clear, and strip out some unnecessary handling in other_select()
since this is a no op everywhere else.

Remove:
- slurm_free_block_info_msg()
- slurm_init_update_block_msg()
- slurm_update_block()

Leave a stubbed version of update_block_msg_t for now.

Removed
- slurm_print_block_info_msg()
- slurm_print_block_info()
- slurm_sprint_block_info()
- slurm_get_select_jobinfo()

Remove select_{g,p}_pack_select_info as well.

Update man page as well. Bump copyright on man page while here.

Remove from step_mgr.c as well - it's a no op on all non-BG systems.

select_g_ba_init() in select/alps still needed for now,
but it does not allocate any dynamic memory and thus
the fini() call can be removed.

sched/backfill: Reset job time limit if needed for deadline scheduling.
bug 5183

- validate more GRES limits
- merge generic parameters
  generic (e.g. "gpu") cpus-per-tres and mem-per-tres values
  will be merged as needed to type-specific (e.g. "gpu:volta")
  specifications.

Do not allow Operators to set this at submit time.

Bug 5306.

I do not see a use for this syntax, especially given that it appends
an extra comma in between the two halves. Only allow the full string
to change to put this in line with the Comment handling.

Remove special handling of an identical AdminComment as well,
since the end result is unchanged, and this avoids a potentially
expensive xstrcmp call.

Bug 5306.

These applied further below, but have been superceeded
by a better commend anyways.

Coverity CID 186551 and 186556

Coverity CID 186552

Identical problem in 3 related functions.
Coverity CID 186550,186553, 186554

Note there was a bug in the configuration parsing logic to get
the partition's DefMemPerGPU (it was stored in the field for
DefCpuPerGPU).

Bug 5286