Proves the phase's headline acceptance criterion: activating tender-radar
for a 2nd tenant triggers zero additional DÖE calls, zero additional cron
jobs (still exactly one 'tender-doe-poll'), and zero additional Tender rows.
Drives the real (unmocked) ModuleRegistryService against a fake prisma to
exercise the genuine activateForTenant() call path.
Passes immediately because Task 2's TenderSchedulerService already
implements the poll-once-fan-out-many invariant correctly — this test
locks in and regression-proofs that already-correct architecture rather
than driving new production code (documented in SUMMARY under TDD Gate
Compliance).
