Remove n_procs for nodes that do not use multiple cores

[tsalo]

Closes none, but addresses an issue identified by @mattcieslak.

Changes proposed in this pull request

• For any single-threaded nodes in any workflow, drop n_procs=omp_nthreads.
• Also drop omp_nthreads=* from any nodes using that. Pretty sure it doesn't do anything.
• Create a WBCommand base interface that sets OMP_NUM_THREADS.
• Use the MultiProc plugin for integration tests, which speeds things up by a fair amount.

[codecov]

Codecov Report

Attention: Patch coverage is 93.33333% with 4 lines in your changes missing coverage. Please review.

Project coverage is 78.26%. Comparing base (3ba1a93) to head (1f30de7).

Files with missing lines	Patch %	Lines
xcp_d/interfaces/workbench.py	92.72%	3 Missing and 1 partial ⚠️

❗ There is a different number of reports uploaded between BASE (3ba1a93) and HEAD (1f30de7). Click for more details.

HEAD has 1 upload less than BASE

Flag	BASE (3ba1a93)	HEAD (1f30de7)
	2	1

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1256      +/-   ##
==========================================
- Coverage   84.10%   78.26%   -5.84%     
==========================================
  Files          59       59              
  Lines        7197     7219      +22     
  Branches     1065     1066       +1     
==========================================
- Hits         6053     5650     -403     
- Misses        880     1296     +416     
- Partials      264      273       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[tsalo]

The integration tests don't seem to contribute to coverage now that we're using the MultiProc plugin, but I think the speed-up is worth the drop in coverage, so I'm going to merge this.