[FIX] Add non-zero floor to T2* values

[tsalo]

Closes #513.

Todo:

• Write test

Changes proposed in this pull request:

• Add a new function to determine smallest non-zero T2* value that will not resolve to zero in optimal combination step. Replace values in the T2* maps > 0 and < this value with the value.
• Add a new unit test for the function. This test doesn't improve coverage, but is important.

[codecov]

Codecov Report

Merging #585 into master will increase coverage by 0.09%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #585      +/-   ##
==========================================
+ Coverage   93.44%   93.53%   +0.09%     
==========================================
  Files          26       26              
  Lines        1953     1965      +12     
==========================================
+ Hits         1825     1838      +13     
+ Misses        128      127       -1     

Impacted Files	Coverage Δ
tedana/decay.py	95.48% <100.00%> (+1.27%)	⬆️

---

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 ddaa4d5...dbfb63c. Read the comment docs.

[jbteves]

LGTM, thanks for adding this fix @tsalo !

[dowdlelt]

This looks good to me, that error is always concerning to see. In the 'appearance is everything' sense, fixing this is a huge deal. I'm approving, but a thought about optionally (verbosely?) logging the number of voxels that needed to be fixed could be useful. Lets call that a future request.

[jbteves]

@dowdlelt would you mind writing up an issue for that?