Summary

Document that scalar, vector, and FTS indexes accept nested field paths using dotted notation (e.g. metadata.user_id, image.embedding, nested.text).

Changes

• Added an "Indexing nested fields" section to the scalar index page with a create_scalar_index + where example.
• Added an "Indexing nested vector fields" subsection to the vector index page with a create_index + search example.
• Added an "Indexing nested string fields" section to the FTS index page with a create_fts_index + fts_columns search example.
• Each section calls out that the canonical dotted path is what you pass in, what list_indices() returns under columns, and what you reference in filters and queries.

Context

Upstream change made native index creation resolve columns using Lance field-path semantics instead of top-level Arrow lookup, and dropped the Python FTS guard that previously rejected dotted paths. Scalar, vector, and FTS indexes now share the same nested-field contract, which wasn't reflected anywhere in the docs.