Rename internal indices to mask

[kklein]

Various MetaLearners use indices to indicate which data point has seen which treatment variant in a field called _treatment_variant_indices, e.g.

metalearners/metalearners/metalearner.py

Line 1339 in 241d681

self._treatment_variants_indices: list[np.ndarray] | None = None

metalearners/metalearners/xlearner.py

Line 106 in 241d681

self._treatment_variants_indices.append(w == treatment_variant)

This naming could suggest that this is a list of indices (e.g. integers) where a specific treatment has been seen. In reality, it is a list of as many booleans as observations. Therefore, calling it a 'mask' might be more appropriate and telling.