feat(limma): implement robust eBayes variance moderation path

[afvallejo]

feat(limma): implement robust eBayes variance moderation path

Description and Context

• JIRA: N/A
• This PR enables the robust eBayes variance moderation path in limma by replacing the previous NotImplementedError behavior when robust=True.
• squeezeVar now uses robust hyperparameter estimation (via robust fitFDist flow) instead of failing.
• The non-robust path (robust=False) remains unchanged to preserve backward compatibility.
• Minor repo hygiene included:
  • add .DS_Store to .gitignore
  • add pytest in pyproject.toml test extra

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• No code (non-breaking change that does not impact code: formatting, build system, documentation...)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Release (version number increment, possibly with documentation updates)

Test strategy

• Verified the robust branch no longer raises NotImplementedError.
• Verified default non-robust behavior is unchanged.
• Recommend running full test suite in CI before merge.

[afvallejo]

Implemented robust eBayes variance moderation in squeezeVar by replacing the previous NotImplementedError robust branch with the robust fitting path, while preserving existing behavior when robust=False. This is a focused, backward-compatible change that unblocks robust limma workflows