feat(eqtl): add preprocessing dag

[ireneisdoomed]

This PR includes:

• new DAG of steps in order to process the whole preprocessing of eQTL Catalog
• redefined eqtl_catalogue step to eqtl_catalogue_ingestion. This is the script that produces a study index and harmonises their raw summary statistics.

When running the step, I hit a bottleneck at the eqtl_catalogue_ingestion level and the job was cancelled.
I want to test it with the recent fixes introduced in #369, it might work now

[codecov-commenter]

Codecov Report

Merging #366 (5dbebce) into dev (42b366c) will increase coverage by 0.24%.
Report is 20 commits behind head on dev.
The diff coverage is 91.30%.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #366      +/-   ##
==========================================
+ Coverage   85.67%   85.92%   +0.24%     
==========================================
  Files          89       90       +1     
  Lines        2101     2117      +16     
==========================================
+ Hits         1800     1819      +19     
+ Misses        301      298       -3     

Files	Coverage Δ
src/airflow/dags/eqtl_catalogue_preprocess.py	100.00% <100.00%> (ø)
src/otg/dataset/l2g_feature_matrix.py	82.92% <ø> (+7.31%)	⬆️
src/otg/eqtl_catalogue_ingestion.py	61.90% <100.00%> (ø)
src/otg/method/l2g/evaluator.py	38.70% <100.00%> (ø)
src/otg/method/l2g/model.py	58.33% <50.00%> (+0.50%)	⬆️

[ireneisdoomed]

I don't think the changes here are relevant any more.

• The plan is to define credible sets from eQTL Catalogue fine mapping results directly, and not from fine mapping summary stats for PICS.
• This PR had also a technical limitation in terms of how the inputs were handled in the DAG, so I never managed to do a full run.

Since we don't have any business logic here and we only touch the Airflow layer, let's just close this.