#4433 perf: drain delivery backlog and copy backups without compaction

[!TIP]
PR walkthrough: Open the visual walkthrough

Large delivery backlogs were throttled by per-batch and per-pass sleeps, while upgrade backups rebuilt the entire SQLite database. Cleanup now runs continuously while it makes progress, and backup-only operations use incremental native page copying while retaining existing verification gates.

• Small transactions, cancellation, maintenance admission, idle waiting and error backoff remain in place. Eligibility and recovery proofs are unchanged.
• Explicit physical compaction still uses VACUUM INTO; page-copy backups preserve free pages.

Validation

• Race tests passed for db, persistence, backups, database, maintenance, inventoryrepair and deliveryretention.
• Historical sweep/protected-gap regressions and persistence store conformance passed.
• Targeted golangci-lint: 0 issues; SQL guard passed.
• Documentation, specification, harness and PR coverage checks passed.
• New regressions reproduced rebuilding and premature waiting before the fix. Fixtures are small; no million-row test.
• CI remediation: worker integration fixtures passed under the race detector; recovery browser failures reproduced, then all 4 affected desktop tests and the mobile recovery test passed with retries disabled.

Checklist

• If I do not have repository write access and this is a large architectural change, I discussed the direction in a linked issue before opening this PR.
• This PR contains one logical change; unrelated work is split into separate PRs.
• I have performed a self-review of my code.
• I have manually tested my changes and they work as expected.
• My changes have tests that cover the new functionality and edge cases.
• If my change touches UI files (apps/web/), I have added or updated Playwright e2e tests in apps/web/e2e/ and verified them with make test-e2e.
• I checked whether this affects public docs in docs/public/** and updated them or noted why no docs change is needed.