ENH: Dilate fmap and bold masks during coregistration

[psadil]

Closes #461

The chosen radius (5 voxels) is a bit of a guess based on what has worked in the dataset where I was noticing failures. Input on a more systematic way to select this would be welcome.

[psadil]

Sorry, but it's not clear to me why that test is failing. I guess it's this doctest?

sdcflows/sdcflows/utils/wrangler.py

Lines 107 to 113 in 1350d57

	>>> find_estimators(
	... layout=layouts['ds000054'],
	... subject="100185",
	... fmapless=False,
	... ) # doctest: +ELLIPSIS
	[FieldmapEstimation(sources=<3 files>, method=<EstimatorType.PHASEDIFF: 3>,
	bids_id='auto_00000')]

[effigies]

Ah, it's global state and parallel tests interacting poorly. I'll figure out what's needed to isolate the state better.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 83.68%. Comparing base (18e5e95) to head (85e0197).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #463      +/-   ##
==========================================
+ Coverage   83.66%   83.68%   +0.01%     
==========================================
  Files          32       32              
  Lines        2816     2819       +3     
  Branches      377      377              
==========================================
+ Hits         2356     2359       +3     
  Misses        390      390              
  Partials       70       70              

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