When using the Managed Data Lake destination, important CDC metadata available from the source is not currently exposed in the final dataset. In particular, fields such as transaction ID and commit sequence/order are often captured in the source (eg. via WAL or binlog) but appear to be discarded or abstracted away in the Iceberg output.

Having the option to passthrough or surface this metadata would be extremely valuable for certain analytic and auditing scenarios. For example:
- Reconstructing which changes occurred as part of a single logical transaction spanning multiple tables
- Performing audit trail analysis for compliance or debugging purposes
- Performing usage and change pattern analysis (eg. frequency or size of multi-table edits)

Ideally this could be implemented as metadata columns (eg. _fivetran_txid, _fivetran_commit_seq) that users could enable per-destination or per-table.