ENH: Add MRIQC's custom multiproc runner that allows passing the pool

[oesteban]

This is useful to minimize the workers' fingerprint.

[effigies]

Is there a reason not to contribute this upstream to nipype?

[codecov]

Codecov Report

Attention: 263 lines in your changes are missing coverage. Please review.

Comparison is base (16609f2) 63.36% compared to head (91aa5d9) 61.05%.
Report is 2 commits behind head on master.

Files	Patch %	Lines
niworkflows/engine/plugin.py	13.48%	263 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #841      +/-   ##
==========================================
- Coverage   63.36%   61.05%   -2.32%     
==========================================
  Files          53       54       +1     
  Lines        6238     6542     +304     
  Branches      907      952      +45     
==========================================
+ Hits         3953     3994      +41     
- Misses       2099     2362     +263     
  Partials      186      186              

Flag	Coverage Δ
reportlettests	∅ <ø> (∅)
unittests	∅ <ø> (∅)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[oesteban]

Is there a reason not to contribute this upstream to nipype?

None - just my laziness and reducing the maintenance effort of nipype.

There's a quasi-reason in the sense that I had the plan to write a plugin based on asyncio and test it with MRIQC, and this was like the first tiny test of a drop-in plugin implementation.

Can we merge this in here to speed up the deployment of sdcflows (which uses this, so we can test), and at the same time I'll send a PR to nipype?

[effigies]

Sure