backup: reject incompatible online cluster restores

[shashi1973]

Summary

Online cluster restore cannot rewrite descriptor IDs when backup SSTs retain their table prefixes.

Backups created with include_all_virtual_clusters use ElidePrefix_Tenant. This strips the tenant prefix but leaves the original table ID in each stored key. Online restore can rewrite the destination span and apply a synthetic tenant prefix, but linking the external SST does not rewrite the table ID embedded in every key.

The compatibility check previously accepted ElidePrefix_Tenant unconditionally. This could link data under incorrect keys, leave restored system-tenant tables empty, and later cause a panic while restoring system tables.

This change treats ElidePrefix_Tenant like ElidePrefix_None when descriptor IDs must change. The restore is rejected during planning with a FeatureNotSupported error, before incompatible SSTs are linked.

The existing successful online cluster restore coverage is preserved. A separate regression scenario creates a secondary tenant, backs up with include_all_virtual_clusters, and verifies that online cluster restore fails cleanly instead of panicking.

Fixes #174659

Tests

• ./dev test pkg/backup -f=TestDataDriven_online_restore_cluster -v
• ./dev test pkg/backup -f=TestDataDriven_online_restore_prefix_backups -v

Release note (bug fix): Online cluster restores now reject backups created with include_all_virtual_clusters when descriptor ID rewrites are required. This prevents system-tenant tables from appearing empty and avoids a panic during system table restoration.

[blathers-crl]

Thank you for contributing to CockroachDB. Please ensure you have followed the guidelines for creating a PR.

My owl senses detect your PR is good for review. Please keep an eye out for any test failures in CI.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[shashi1973]

Hi @kev-cao,

I wanted to follow up on this PR when you have an opportunity. Could you please review the implementation and let me know whether it adequately addresses the linked issue?

The implementation and relevant regression coverage are included, and the applicable tests pass locally. I would be glad to make any further changes or add additional test coverage based on your feedback.

Thank you for your time and review.