Add relay-status CLI subcommand (closes #4)

[spcpza]

Summary

Implements lightning-memory relay-status as requested in issue #4.

Usage

# Human-readable:
$ lightning-memory relay-status
Lightning Memory — relay status
DB: /home/user/.lightning-memory/memory.db

  ✓  wss://relay.damus.io  (142ms)
  ✓  wss://nos.lol  (89ms)
  ✗  wss://relay.nostr.band
     connection refused

  memories in DB : 47
  events pushed  : 44
  last push      : 2026-03-08 14:22:01
  last pull      : 2026-03-08 13:00:15

2/3 relays reachable.

# Machine-readable (for scripts and agents):
$ lightning-memory relay-status --json
{
  "relays": [
    {"url": "wss://relay.damus.io", "reachable": true, "latency_ms": 142, "message": "ok"},
    ...
  ],
  "sync": {"total_memories": 47, "pushed_events": 44, ...},
  "db_path": "/home/user/.lightning-memory/memory.db"
}

Implementation

lightning_memory/cli.py (new file):

• _probe_relay() — sends a limit:0 REQ and waits for EOSE, the lightest possible liveness check that exercises the full WS + NIP-01 handshake. All relays probed concurrently.
• _sync_stats() — reads push/pull counts and timestamps from the local SQLite sync_log / sync_cursor tables.
• _relay_status_async() — combines both, prints human or JSON output, returns exit code 0 (all OK) or 1 (any relay unreachable).
• cmd_relay_status() — sync wrapper for the entry point.

lightning_memory/server.py (minor patch):

• main() now checks sys.argv for a relay-status subcommand before starting the MCP server. No new binary needed — the existing lightning-memory entry point gains the new command.

tests/test_relay_status.py (new file): 15 tests covering all branches — reachable relay, unreachable relay, NOTICE error, latency measurement, DB stats (populated / missing / empty), human output, JSON output, sync stats display, and server dispatch.

Exit codes

Code	Meaning
0	All configured relays reachable
1	One or more relays unreachable

---

If this is useful, sats tips are welcome: ⚡ sensiblefield821792@getalby.com