fix(rust_brain): preserve HLC on snapshot restore and gossip receive

[cursor]

Bug and impact

Critical silent data loss in distributed deployments after snapshot restore.

v0.6.0 added Hybrid Logical Clock (HLC) monotonic enforcement on writes, but restore_from_file() and gossip.receive() did not preserve or apply HLC timestamps:

1. After snapshot restore: restored nodes received fresh _hlc.now() timestamps instead of their saved HLC. Valid causal successor writes (e.g. gossip replay with hlc=(5000, 11, "nodeA")) were rejected as stale, causing silent data loss.
2. Gossip without HLC: events missing hlc could overwrite fresher local data because remember() generated a new timestamp.
3. Gossip with stale HLC: when hlc was present in the event but not passed through, stale remote values could overwrite fresher local data.

Root cause

restore_from_file() constructed MemoryNode without the hlc field from the snapshot. bulk_write() and gossip.receive() also omitted HLC passthrough, breaking causal ordering introduced in v0.6.0.

Fix

• Add _parse_hlc() for snapshot/wire normalisation (with legacy ts_ns fallback for pre-v0.6.0 snapshots)
• restore_from_file(): restore hlc per node and advance the global _hlc
• bulk_write(): pass through hlc from row payloads
• gossip.receive(): apply hlc, call update_hlc(), reject stale updates, skip missing-HLC overwrites on existing keys

Validation

• New regression tests in test_enterprise_backup.py and test_gossip.py
• Full suite: 192 passed, 8 skipped