Handle a None encoding in load_data when assertions are disabled

[arpitjain099]

Proposed changes

Follow-up to #978, taking @jeongyoonlee's note from that review.

OneHotEncoder.transform signals "no column produced a non-empty encoding" with an assert, and load_data catches it. Under python -O (or PYTHONOPTIMIZE=1) that assertion is compiled out, so transform returns None instead of raising, the except never runs, and .toarray() lands on None:

$ python -O -c "from causalml.features import load_data; ..."
AttributeError: 'NoneType' object has no attribute 'toarray'

That is the same constant-categorical input the assertion path already handles, so it works normally and fails only when assertions are off.

The fix keeps fit_transform in the try and moves .toarray() out behind a None check, so both signals converge on the same branch. No re-deriving of the encoder's own rule.

Types of changes

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation Update (if none of the other choices apply)

Checklist

• I have read the CONTRIBUTING doc
• I have signed the CLA
• Lint and unit tests pass locally with my changes
• I have added tests that prove my fix is effective or that my feature works
• I have added necessary documentation (if appropriate)
• Any dependent changes have been merged and published in downstream modules

Further comments

The test monkeypatches OneHotEncoder.fit_transform to return None, which reproduces the optimized-mode behavior deterministically without spawning a second interpreter. It fails on current master with the AttributeError above and passes with this change.

I also ran the real thing both ways. On master, python -O raises; with this patch it returns a (7, 1) array. pytest tests/test_features.py is 7 passed and black --check is clean.

[jeongyoonlee]

LGTM