Speed up identity distance metric computation for two different sequence arrays

[felixpetschko]

Summary

This PR speeds up the identity distance metric computation when comparing two different sequence arrays via the IdentityDistanceCalculator.

Instead of checking every pairwise combination of seqs and seqs2, the implementation now builds a lookup from sequences in seqs2 to their column indices and uses it to get matching matrix entries directly.

For example, the previous performance issue became a problem when running the following call with a large number of cells:

ir.pp.ir_dist(mdata, vdjdb, metric="identity", sequence="aa")

With this change, I was able to run the IdentityDistanceCalculator with millions of sequences in seconds.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 11 lines in your changes missing coverage. Please review.
✅ Project coverage is 19.75%. Comparing base (05344e5) to head (644ad8b).

Files with missing lines	Patch %	Lines
src/scirpy/ir_dist/metrics.py	0.00%	11 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #701      +/-   ##
==========================================
- Coverage   19.76%   19.75%   -0.02%     
==========================================
  Files          51       51              
  Lines        4578     4581       +3     
==========================================
  Hits          905      905              
- Misses       3673     3676       +3     

Files with missing lines	Coverage Δ
src/scirpy/ir_dist/metrics.py	14.56% <0.00%> (-0.08%)	⬇️

🚀 New features to boost your workflow:

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

[grst]

Thanks!