Enforce HANA cascade-constraint deploy before draft_admin_integrity serves (per-env ordering) #2968

Context

Follow-up to #2909 (PR #2967), which enabled cds.fiori.draft_admin_integrity: true (CAP 10.2 optimized draft deletion). The flag is committed ON, so the next deploy to any environment serves with it on.

The risk (deferred minor from the #2909 whole-branch review)

Per-env safety currently relies on the deployer landing the generated HANA ON DELETE CASCADE constraints (.hdbconstraint, NOT VALIDATED) before the flag serves in that environment. Nothing in code enforces this ordering.

If the runtime serves flag-on before the cascade constraints exist on HANA, draft activation/root-cancellation delete only the DraftAdministrativeData record and the DB does not cascade the child draft rows → orphaned draft rows accumulate in the draft tables.

• Impact: draft-table cruft, not active-content data loss.
• Mitigations already in place: the migration runbook (docs/developers/reference/tutorials-ims-gotchas.md → "Optimized draft deletion"), the NOT VALIDATED constraint deploy (won't be blocked by pre-existing orphans), and the DEV → test → PROD no-main-hotfix model.

Proposed work

Make the ordering safe without relying on deployer discipline. Options to evaluate:

1. A deploy-time guard / preflight check that verifies the cascade constraints exist on the target HANA schema before (or as part of) flipping the serving flag.
2. A one-shot orphan-cleanup pass using cds.fiori.draft_orphan_cleanup: true wired into the migration, then disabled.
3. A post-deploy smoke/probe that detects orphaned draft rows and alerts.

References

• Issue Use new Draft Delete - cds.fiori.draft_admin_integrity: true in CAP #2909, PR feat(cap): adopt draft_admin_integrity (optimized draft deletion) #2909 #2967
• Runbook: docs/developers/reference/tutorials-ims-gotchas.md § "Optimized draft deletion — draft_admin_integrity"
• CAP 10.2 release notes: https://cap.cloud.sap/docs/releases/2026/oct26#optimized-draft-deletion