Speed up test DBs and reduce memory footprint

[stephanos]

Combination of independent experiments validated on parallel feature/tweak-db-* branches under standard ARM runners. Per-job memory peaks were captured by the upgraded memory_monitor.sh.

Changes:

• Healthcheck interval 3s -> 1s (cheap; fast services like Postgres no longer wait an extra tick).
• Cassandra:
  • MAX_HEAP_SIZE 2G -> 1G (HEAP_NEWSIZE 200M -> 100M). Per-job Cassandra container peak: 2586MB -> 1533MB (-41%). No runtime regression on cass_* jobs; cass_es8 -10% on test runtime.
  • JVM flags to skip the ~30s gossip-settle wait that's pointless on a single-node test cluster (ring_delay_ms=1000, skip_wait_for_gossip_to_settle, single-token settings).
• Elasticsearch / OpenSearch heap 1G -> 512M. Per-job ES7 peak 1404MB -> 852MB (-40%); ES8 1464MB -> 943MB (-36%). System peak on ES-using jobs dropped 13.9GB -> 11.6GB median (-17%). No runtime regression observed.
• Postgres fsync=off, synchronous_commit=off, full_page_writes=off; MySQL --innodb-flush-log-at-trx-commit=0, --innodb-doublewrite=0, --skip-log-bin. CI databases are ephemeral so durability has no value but fsync overhead is real. Test runtime: mysql8 -23%, postgres12 -26%.
• memory_monitor.sh now tracks per-container memory via docker stats and prints a system+per-container peak summary at run end (visible at the bottom of every CI job log).

Combined wall-clock saving on shard jobs alone: ~2000s across the full matrix. Stability either equal or improved (5/40 failures vs 10/43 baseline on the standard runners; 8-core runners should clear those timeout-driven retries).

What changed?

Describe what has changed in this PR.

Why?

Tell your future self why have you made these changes.

How did you test it?

• built
• run locally and tested manually
• covered by existing tests
• added new unit test(s)
• added new functional test(s)

Potential risks

Any change is risky. Identify all risks you are aware of. If none, remove this section.