Reorder layers in adaptive mask plot

[tsalo]

Closes none, but spins off a small bugfix from #1478.

Changes proposed in this pull request:

• Reverse order of mask lines in plot_adaptive_mask so that initial mask is on the bottom and the classification mask is on the top.

[tsalo]

Here are the figures from the five-echo test. You'll be able to see that the majority of the initial mask is retained in the final mask, and the new plot correctly shows a blue line (Classification, OC & Initial) for that mask instead of a red line (Initial mask only).

Before

After

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.43%. Comparing base (fd18a66) to head (edc89aa).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1480   +/-   ##
=======================================
  Coverage   90.43%   90.43%           
=======================================
  Files          31       31           
  Lines        5100     5100           
  Branches      937      937           
=======================================
  Hits         4612     4612           
  Misses        326      326           
  Partials      162      162           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[eurunuela]

LGTM! Thanks!

[handwerkerd]

@tsalo I was traveling, but I wanted to mention we did originally compare the two color contours (see : #1333 (comment) )

The reason I prefer the initial mask as the top layer is because that makes the user-provided mask very clear and highlights any deviations from the initial mask.

That is, if you look at the following, it's clear that a non-trivial number of voxels in the user-provided mask were not included in the ICA mask.

For the flipped version, there's a little dot of magenta but it seems to be a bit less clear how much is lost.

If others really like the now-current flipped ordering, I won't push back more than this.

[tsalo]

I can revert the change, but we should fix the labels. The "initial mask only" label is accurate with the new order because any time you see magenta it means that that bit corresponds to voxels that were only present in the initial mask. In the old version that label wasn't accurate, since it reflected the initial mask, but not only the initial mask.