sql: TTL Job Maximum Execution Time Parameter

Today admins are unable to bound TTL job execution to a defined maintenance window. Not doing this means there is no guarantee that TTL operations will stop at the end of a scheduled window, risking uncontrolled impact to production workloads outside of approved maintenance periods.

Requesting to introduce per-table or cluster-level setting to cap TTL job execution time(ttl_max_runtime). When the limit is reached, the TTL job should stop cleanly and run again on the next scheduled run starting from where it left off. Default of ttl_max_runtime = 0 indicates there is no timelimit on TTL jobs. Any non-zero positive value (e.g. '2h', '4h') indicates the execution time limit.

Jira issue: CRDB-64335