Support cluster-level Monte Carlo FWE correction in the MKDAChi2 Estimator

[tsalo]

Closes None.

Changes proposed in this pull request:

• Implement cluster-level FWE correction in the MKDAChi2 estimator.
• Fix voxel-level FWE correction in the MKDAChi2 estimator.
  • It looks like the FWE null distributions really shouldn't be treated as two-sided tests, which is how the vFWE was being done. The FWE null distribution reflect maximum values across both negative and positive values/clusters, so we need to take the absolute values and compare those in a one-sided way to the null distributions before re-signing them.

[codecov]

Codecov Report

Merging #650 (874bff6) into main (02387b9) will increase coverage by 0.22%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #650      +/-   ##
==========================================
+ Coverage   85.02%   85.24%   +0.22%     
==========================================
  Files          40       40              
  Lines        4460     4528      +68     
==========================================
+ Hits         3792     3860      +68     
  Misses        668      668              

Impacted Files	Coverage Δ
nimare/meta/cbma/base.py	94.94% <100.00%> (+0.22%)	⬆️
nimare/meta/cbma/mkda.py	97.09% <100.00%> (+0.65%)	⬆️

---

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 02387b9...874bff6. Read the comment docs.

[tsalo]

For some reason the threshold parameter isn't having any effect on the plot_stat_map figures, but I don't think it's a problem for the new feature.

[tsalo]

I may have to revisit this based on #655 (comment). Now in #663.