Harden DB writes against partial table loss

[jeanconn]

Description

Harden DB writes against partial table loss.

These issues kept popping up in local repro work so collected some solutions.

• make DB writes resilient to partial failures so one bad write does not silently
  destroy table state
• fix region-id allocation so add_regions() does not reuse IDs when
  last_region_id is absent, stale, or behind the current table contents
• repair duplicate region_id values during region-table sync while keeping
  destination IDs stable where possible
• trim oversized region snapshot coverage into a more focused dedicated test module

Interface Impacts

• DB write failures should now be more explicit and less likely to leave tables in a
  partially replaced state
• region tables should no longer hand out duplicate numeric IDs or reissue
  retired ones during add/sync operations

Testing

• expands DB unit coverage for partial-write recovery and adds focused region tests
  for region_id allocation, stale counters, and sync-time duplicate repair

Unit tests

• No unit tests
• Mac
• Linux
• Windows

Independent check of unit tests by [REVIEWER NAME]

• [PLATFORM]:

Functional tests

No functional testing.