Force finufft signals to be complex

[garrettwrong]

This should work around the changes put in FINUFFT v2.0.4. Also it seems like their project lead might prefer to enforce this instead of casting in the future.

[garrettwrong]

Closes #572

Deprecates #573

[garrettwrong]

See flatironinstitute/finufft#203

[codecov]

Codecov Report

Merging #575 (3dcfaeb) into develop (ef91777) will decrease coverage by 0.22%.
The diff coverage is 84.61%.

@@             Coverage Diff             @@
##           develop     #575      +/-   ##
===========================================
- Coverage    87.42%   87.20%   -0.23%     
===========================================
  Files          108      108              
  Lines         8172     8000     -172     
===========================================
- Hits          7144     6976     -168     
+ Misses        1028     1024       -4     

Impacted Files	Coverage Δ
src/aspire/commands/preprocess.py	0.00% <0.00%> (ø)
src/aspire/__init__.py	100.00% <100.00%> (ø)
src/aspire/basis/ffb_3d.py	100.00% <100.00%> (ø)
src/aspire/commands/extract_particles.py	85.71% <100.00%> (+5.71%)	⬆️
src/aspire/nufft/finufft.py	100.00% <100.00%> (ø)
src/aspire/source/image.py	95.43% <100.00%> (-0.28%)	⬇️
src/aspire/source/simulation.py	99.38% <100.00%> (-0.20%)	⬇️

---

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 d5d7747...3dcfaeb. Read the comment docs.

[janden]

So do I understand correctly in that 2.0.3 allowed real dtypes and did the cast to complex, but that 2.0.4 did not cast (but allowed real dtypes, hence the bug)?

[garrettwrong]

Yes, that is correct. Prior versions of FINUFFT cast real signals to complex for the user.