fix: support multi-series X for aligner fit

[Saswatsusmoy]

Reference Issues/PRs

Fixes #491

What does this implement/fix? Explain your changes.

Aligners need X as a panel / df-list (list of series). fit only resolved a single series per X_* / y_* slot and routed object_type=aligner through the forecaster-style fit(y, X=...) path. That produced the df-list type error on a single series (e.g. airline) and a signature clash when only X was set.

This change:

• lets X_handle / X_dataset be a string or a list of ids
• assembles a df-list for list inputs (Series → one-column DataFrame)
• routes aligners to fit(X) with a clear error on a single series/table
• extends call_method *_dataset / *_data_handle injection the same way for multi-series lists

Shared resolve is used by sync fit and fit_async. Tests use AlignerNaive (no soft deps).

Does your contribution introduce a new dependency? If yes, which one?

No.

What should a reviewer concentrate their feedback on?

• List resolve for X_handle / X_dataset and Series → DataFrame for df-list items
• Aligner branch in Executor.fit (including single flat DataFrame vs MultiIndex panel)
• call_method list injection parity with fit (len >= 2)

Any other comments?

make check is green locally (206 tests). Commit-time pre-commit passed on this branch.

PR checklist

For all contributions

• I've added unit tests and made sure they pass locally (make check).
• I've added the tool to the online documentation in docs/source/. n/a (schema/description only on the existing fit tool; no new tool)
• I've updated the existing example scripts or provided a new one to showcase how my tool works in examples/. n/a

[Saswatsusmoy]

one caveat for reviewers / users: a single series (e.g. y_dataset=airline alone) still fails, on purpose. aligners need multiple series, so we return a short error asking for X_handle or X_dataset as a list of at least two ids. the bug was never being able to pass that list and wrong fit routing. multi-series list path is what this pr fixes.