feat: make the gene list of single-view sklearn models configurable

[jfrog64]

PR Checklist for all PRs

• This comment contains a description of changes (with reason)
• Referenced issue is linked — no existing issue, came up while benchmarking baselines on other gene spaces
• If you've fixed a bug or added code that should be tested, add tests!
• Documentation in docs is updated — no new module and no signature that the API docs render, so nothing to add

Changes

New features

load_single_cell_line_view hard-codes the gene list it subsets gene_expression with:

if "gene_expression" in cell_line_views:
    return load_and_select_gene_features(
        feature_type="gene_expression",
        gene_list="landmark_genes_reduced",
        ...
    )

So every single-view sklearn baseline — ElasticNet, RandomForest, SVR, SingleDrug*, … — is
locked to the landmark genes. The other gene lists that ship with the framework
(drug_target_genes_all_drugs, gene_list=None for the full matrix) are reachable for hand-written
models, but not for the baselines, even though the baselines are exactly where one wants to vary the
feature space to see how much of a model's performance comes from the gene selection rather than the
learner.

This makes it configurable in two steps, without changing any default:

1. load_single_cell_line_view gains gene_list: str | None = "landmark_genes_reduced" and forwards it.
   The default is the previously hard-coded value, so every existing caller behaves exactly as before.

2. SklearnModel gains a gene_list class attribute with the same default, and build_model reads it
   from the hyperparameters:

   self.gene_list = hyperparameters.get("gene_list", type(self).gene_list)

   This is the same pattern the class already uses for cell_line_views and drug_views one line above.
   The value stays in self.hyperparameters, so save()/load() carry it and predict() uses the same
   gene space the model was trained on.

gene_list is deliberately not added to baselines/hyperparameters.yaml, so the tuning grids are
untouched and nothing changes for existing runs. It becomes available to anyone who passes a custom
hyperparameter set.

Tests: tests/test_sklearn_gene_list.py builds a temporary meta/gene_lists directory with two lists
and covers

• omitting gene_list reproduces the previously hard-coded landmark_genes_reduced behaviour
• an explicit list is honoured
• gene_list=None loads the full expression matrix
• the SklearnModel default equals the class attribute
• the "gene_list" hyperparameter reaches load_cell_line_features and lands in self.hyperparameters
• setting it on one instance does not leak into the class attribute or into other instances

Bug fixes

Maintenance

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.10%. Comparing base (7d24cd6) to head (3269d9f).
⚠️ Report is 75 commits behind head on development.
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@               Coverage Diff               @@
##           development     #464      +/-   ##
===============================================
+ Coverage        80.34%   82.10%   +1.75%     
===============================================
  Files              101      121      +20     
  Lines             8171     9833    +1662     
===============================================
+ Hits              6565     8073    +1508     
- Misses            1606     1760     +154     

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

🚀 New features to boost your workflow:

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