ua: cluster ID mismatch during UA migration

**Is your feature request related to a problem? Please describe.**
During UA migration experiment, we noticed an issue in issue when the cluster is in mixed mode. Mixed mode refers to the cluster state during rolling update where some nodes in the cluster are configured to treat TenantOne as system tenant while the rest are configured to treat TenantTwo as system tenant.

Cluster IDs of a secondary tenant is derived using storage cluster ID and tenant ID: https://github.com/cockroachdb/cockroach/blob/c0442bbfd9c65da745294959ab76bffbfe738690/pkg/rpc/context.go#L502-L524

During rolling update, TenantOne transitions from being a system tenant to secondary tenant. Any operations like backup started before the update fail with cluster ID mismatch error.

**Describe alternatives you've considered**
One option is to use derived cluster ID for all tenant system or secondary but rolling out this change can be problematic. Another option is to keep the cluster ID unchanged for TenantOne.

Jira issue: CRDB-45295