RF: Refactor neuropil.separate

[scottclowe]

• Remove duplicated code by refactoring so we have common logic only once
• By generalising the code, we enable working with arbitrary sklearn.decomposition classes (e.g. FactorAnalysis), though this feature is undocumented.
• Documentation is improved.

@swkeemink Would be great if you could double-check my descriptions of the methodology for normalising the mixing matrix to rank the signals.

[codecov]

Codecov Report

Merging #188 (d89a3d6) into master (7a4dcf5) will increase coverage by 1.45%.
The diff coverage is 97.05%.

❗ Current head d89a3d6 differs from pull request most recent head 78385ee. Consider uploading reports for the commit 78385ee to get more accurate results

@@            Coverage Diff             @@
##           master     #188      +/-   ##
==========================================
+ Coverage   92.29%   93.75%   +1.45%     
==========================================
  Files           8        8              
  Lines         935      912      -23     
  Branches      192      188       -4     
==========================================
- Hits          863      855       -8     
+ Misses         37       30       -7     
+ Partials       35       27       -8     

Flag	Coverage Δ
nbsmoke	66.99% <35.29%> (?)
unittests	93.42% <97.05%> (+1.22%)	⬆️

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

Impacted Files	Coverage Δ
fissa/neuropil.py	98.18% <97.05%> (+6.75%)	⬆️
fissa/roitools.py	88.98% <0.00%> (-1.25%)	⬇️
fissa/extraction.py	95.86% <0.00%> (+0.82%)	⬆️
fissa/core.py	97.12% <0.00%> (+2.88%)	⬆️

---

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 7a4dcf5...78385ee. Read the comment docs.