[REF] Split eigendecomp into ICA and PCA files

[tsalo]

Closes None. References #166. This was originally part of #247, but has been split off because it is self-contained and should be merged separately.

Changes proposed in this pull request:

• Split eigendecomp into ICA and PCA files.
• Move kundu_tedpca into selection module.
• Add out_dir argument to tedpca.
• Remove pcastate file. PCA doesn't take very long.
• Transpose PCA mixing matrix internally so it's in the same format as the ICA mixing matrix.

[codecov]

Codecov Report

Merging #265 into master will increase coverage by 0.66%.
The diff coverage is 17.96%.

@@            Coverage Diff             @@
##           master     #265      +/-   ##
==========================================
+ Coverage   46.08%   46.75%   +0.66%     
==========================================
  Files          33       35       +2     
  Lines        2044     2034      -10     
==========================================
+ Hits          942      951       +9     
+ Misses       1102     1083      -19

Impacted Files	Coverage Δ
tedana/workflows/tedana.py	14.64% <0%> (ø)	⬆️
tedana/selection/__init__.py	100% <100%> (ø)	⬆️
tedana/decomposition/__init__.py	100% <100%> (ø)	⬆️
tedana/selection/tedpca.py	11.11% <11.11%> (ø)
tedana/decomposition/pca.py	15.49% <15.49%> (ø)
tedana/decomposition/ica.py	26.92% <26.92%> (ø)

---

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 e5cab53...540ae8f. Read the comment docs.

[emdupre]

Sorry, why did we want to put ICA and PCA in their own modules ?

[tsalo]

The other changes are perhaps more important (e.g., moving kundu_tedpca into the selection module and removing the pcastate file), but my thinking is that separating ICA and PCA will work better when we add new decomposition methods (one of our future goals). Plus, if we want to support low-memory incremental PCA, then that may merit an additional function within pca.py (like run_mlepca). If we end up expanding the ICA- and PCA-related functions, then it makes sense to group them within their own submodules.

[emdupre]

This LGTM ! Very small comments 👍

[emdupre]

Should the docstring include that this is z-scored ?

[emdupre]

This LGTM ! ✨