[ENH] Coerce images to 32-bit

[jbteves]

Closes #757 .

Changes proposed in this pull request:

• Downscales 64-bit float, int arrays to 32-bit arrays just before image save
• Adds some notes to the io.save_img docstring

[tsalo]

The changes you've made look good to me. I have one question/recommendation, but I was also thinking maybe we should have something in the outputs page about the change in datatypes?

[codecov]

Codecov Report

Merging #759 (a52af7b) into main (12f16eb) will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #759      +/-   ##
==========================================
+ Coverage   92.83%   92.85%   +0.01%     
==========================================
  Files          27       27              
  Lines        2192     2197       +5     
==========================================
+ Hits         2035     2040       +5     
  Misses        157      157              

Impacted Files	Coverage Δ
tedana/io.py	93.36% <100.00%> (+0.17%)	⬆️

---

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 12f16eb...a52af7b. Read the comment docs.

[jbteves]

Hm, I'm not sure what we'd put in the outputs page. This change should be relatively invisible aside from freer disks 🙂. Programs should read everything correctly, and float32 is the most widely supported data type of them all. I think only developers need to be aware, but maybe there's something I'm not thinking of.

[tsalo]

That's fair. I guess not many people will care. Approving!