ENH: Keep a registry of already-used identifiers (and auto-generate new)

[oesteban]

This is the projection of the B0FieldIdentifier metadata proposed for the BIDS spec. Having an identifier for each estimation strategy is the most robust way of minimizing the computational load and reliably keep track of fieldmap information.

Followed-up-by: #129.
Resolves: #125.

[codecov]

Codecov Report

Merging #130 (b5fdecd) into dev/1.4.0 (5588e11) will increase coverage by 0.10%.
The diff coverage is 92.30%.

@@              Coverage Diff              @@
##           dev/1.4.0     #130      +/-   ##
=============================================
+ Coverage      88.44%   88.55%   +0.10%     
=============================================
  Files             15       15              
  Lines            926      952      +26     
  Branches         109      120      +11     
=============================================
+ Hits             819      843      +24     
- Misses            75       76       +1     
- Partials          32       33       +1     

Flag	Coverage Δ
travis	71.53% <92.30%> (+0.58%)	⬆️
unittests	88.51% <92.30%> (+0.10%)	⬆️

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

Impacted Files	Coverage Δ
sdcflows/utils/phasemanip.py	91.30% <ø> (ø)
sdcflows/workflows/base.py	95.00% <ø> (ø)
sdcflows/fieldmaps.py	83.68% <92.30%> (+1.94%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5588e11...b5fdecd. Read the comment docs.

[oesteban]

Okay, I think this PR would be the most interesting part of the big refactor I'm undertaking on dev/1.4.0.

It also starts with the implementation of the B0FieldIdentifier metadata, if it finally makes it into the spec (if not, there are technical reasons to still keep this - although we cannot count on PyBIDS to do the job in that case).

I think this tracking helps justify the utilization of the FieldmapEstimation object (instead of just a function), which was a concern for @mgxd. The idea is to keep building on the FieldmapEstimation class and make it the factory for the creation of the estimation workflows (i.e., it will be the recommended API to sdcflows.workflows.fit.

Thoughts?

[oesteban]

Okay, I'm going to continue with the overhaul - sorry for not allowing the adequate review window to properly eyeball through all the changes.