MAINT: Revert addition of tox

[oesteban]

Let's see whether the failing test comes from the tox insulation of the environment.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 63.83%. Comparing base (4a6cf72) to head (2cba903).

Additional details and impacted files

@@             Coverage Diff             @@
##             main     #232       +/-   ##
===========================================
+ Coverage   53.35%   63.83%   +10.48%     
===========================================
  Files          17       17               
  Lines         849      849               
  Branches      143      143               
===========================================
+ Hits          453      542       +89     
+ Misses        369      267      -102     
- Partials       27       40       +13     

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

[oesteban]

I'm increasing the threshold for the failing test. At this point, it is clear:

• It's not ANTs (tested with 2.3 and 2.5)
• It's not the environment or tox (it works locally but doesn't on GHA)

I'm afraid we might be hitting the nitransforms bug cc/ @effigies.

[effigies]

The nitransforms bug was related to adding per volume transforms to a chain. Should be fixed in the latest release. Are you able to fix this issue by pinning <24?

[oesteban]

Are you able to fix this issue by pinning <24?

That was about to be my next commit :D

This is where I think we are hitting game over:

eddymotion/test/test_integration.py

Lines 44 to 55 in 4a6cf72

	xfms = nt.linear.LinearTransformsMapping(
	[
	nb.affines.from_matvec(nb.eulerangles.euler2mat(x=0.03, z=0.005), (0.8, 0.2, 0.2)),
	nb.affines.from_matvec(nb.eulerangles.euler2mat(x=0.02, z=0.005), (0.8, 0.2, 0.2)),
	nb.affines.from_matvec(nb.eulerangles.euler2mat(x=0.02, z=0.02), (0.4, 0.2, 0.2)),
	nb.affines.from_matvec(nb.eulerangles.euler2mat(x=-0.02, z=0.02), (0.4, 0.2, 0.2)),
	nb.affines.from_matvec(nb.eulerangles.euler2mat(x=-0.02, z=0.002), (0.0, 0.2, 0.2)),
	nb.affines.from_matvec(nb.eulerangles.euler2mat(y=-0.02, z=0.002), (0.0, 0.2, 0.2)),
	nb.affines.from_matvec(nb.eulerangles.euler2mat(y=-0.01, z=0.002), (0.0, 0.4, 0.2)),
	],
	reference=b0nii,
	)

[oesteban]

It is nitransforms.