Module migration 

Embedding model migration for AI-Lake tables.

Reads all chunks from a table, re-embeds them with a new model, and writes new files with the updated embedding column. Two strategies are supported:

• AtomicReplace: replaces each file one at a time. Lower peak storage, but during the migration window different shards may have different columns.
• DualWriteThenCutover: writes new files containing both old and new columns, then atomically replaces all old files. Higher peak storage, zero downtime.

Structs

MigrationJob

Migrates embedding columns in an AI-Lake table to a new model.

MigrationProgress

Progress reported via on_progress callback.

Enums

MigrationStrategy

How files are replaced during migration.

Type Aliases

EmbedFn

ProgressFn