FEAT: expose explicit data shape metadata in load_data_source

[biru-codeastromer]

Summary

Closes #416.

This adds explicit agent-friendly shape metadata to load_data_source and load_data_source_async responses so MCP clients can reason about loaded data handles without guessing from raw column names and dtypes.

What changed

• add a shared metadata builder in Executor
• expose:
  • target_scitype
  • target_variates
  • has_exog
  • exog_variates
  • index_type
  • n_target_columns
  • n_exog_columns
• apply the same metadata enrichment to both sync and async load paths
• add focused tests for:
  • datetime-indexed target with multivariate exogenous data
  • range-indexed target without exogenous data

Why this matters

This makes loaded data self-describing for agents before they choose estimators or workflow branches. In particular, it reduces ambiguity around:

• whether a handle is univariate or multivariate
• whether exogenous variables are present
• whether the time index is datetime-like or integer/range based

Validation

• .venv/bin/ruff check src/sktime_mcp/runtime/executor.py tests/test_data_sources.py
• .venv/bin/ruff format --check src/sktime_mcp/runtime/executor.py tests/test_data_sources.py
• .venv/bin/pytest tests/test_data_sources.py -q
• .venv/bin/pytest -q

[Shashankss1205]

The idea (self-describing handles) still fits the current design and #416 is open. Two things are needed: (1) rebase onto main; there is one small conflict in executor.py where main now normalises the stored handle to a PeriodIndex (#531): keep that block and add your metadata call after it; (2) compute target_scitype / n_target_columns / target_variates instead of hardcoding 'Series', 1, 'univariate'; inspect_data already uses sktime.datatypes.check_is_scitype, so please reuse that so panel/multivariate handles are reported correctly.

[Me-Priyank]

Had a look, target_variates is derived from y.ndim, but base.py:77 always returns a Series so it's always 1, and target_scitype is hardcoded to "Series". For a MultiIndex panel that comes back as "Series", which may be worse than leaving the field out. inspect_data.py:66-73 already uses check_is_scitype, reusing that would cover panels.

Also index_type is read before the PeriodIndex normalisation at executor.py:1236-1240, so on the default path it reports "datetime" for a handle whose index is actually a PeriodIndex.