feat: LD index and block matrix extraction for a studyLocus

[Daniel-Considine]

Adds get_gnomad_matrix() function to study_locus.py. This function takes a study locus, finds the relevant study ID, extracts the major ancestry, and then returns the ancestry specific hail ld index and block matrices for that window.

[codecov-commenter]

Codecov Report

Attention: 175 lines in your changes are missing coverage. Please review.

Comparison is base (42b366c) 85.67% compared to head (99fbf48) 85.94%.
Report is 115 commits behind head on dev.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #463      +/-   ##
==========================================
+ Coverage   85.67%   85.94%   +0.27%     
==========================================
  Files          89      100      +11     
  Lines        2101     2875     +774     
==========================================
+ Hits         1800     2471     +671     
- Misses        301      404     +103     

Files	Coverage Δ
src/airflow/dags/common_airflow.py	90.38% <100.00%> (ø)
src/airflow/dags/dag_preprocess.py	100.00% <ø> (ø)
src/airflow/dags/finngen_preprocess.py	100.00% <100.00%> (ø)
src/airflow/dags/gwas_curation_update.py	100.00% <100.00%> (ø)
src/gentropy/__init__.py	100.00% <ø> (ø)
src/gentropy/assets/__init__.py	100.00% <ø> (ø)
src/gentropy/assets/data/__init__.py	100.00% <ø> (ø)
src/gentropy/assets/schemas/__init__.py	100.00% <ø> (ø)
src/gentropy/cli.py	91.66% <100.00%> (ø)
src/gentropy/common/Liftover.py	80.64% <ø> (ø)
... and 71 more

... and 38 files with indirect coverage changes

[addramir]

Is it ready for the review?

[Daniel-Considine]

I haven't yet added tests for the functions I have added to study_locus.py

@DSuveges
To get the tests to pass on some of the pairwise_ld.py code I've just hacked it together to get it working, so there is probably some issues here. Do you think the functions I've written would make more sense to be a part of pairwise_ld.py instead?

[DSuveges]

To extract the square NumPy matrix for a random study locus (5_59546948_C_T) using a 1 mega-base window in non-finish European populations takes about 40s running on my laptop.

Nice. I'll review this PR together with the Z-score one today.

[DSuveges]

My only concern is the layout of the locus index retriever function. Which functionality is more or less already in this function: get_ld_variants here

[DSuveges]

OK, I think all makes sense now. In a separate effort we can update the other function that the ld index generation uses.