[discord] expose slashCommandDeploy mode in config (changed-only / disabled)

Environment

• openclaw 2026.4.29
• Discord plugin built-in (5 accounts: secretary, hidev, hidoc, himarketer, hiplanner)

Symptoms

Every gateway restart triggers Discord slash-command redeploy, which hits 429 rate limits in clusters:

[discord] native slash command deploy warning (not message send):
  You are being rate limited. (status=429,
  body={"message":"You are being rate limited.","retry_after":1.192,"global":false})

Observed 5+ such warnings within ~30 seconds of each restart across multiple accounts. Effects:

• Some accounts' slash-commands silently fail to register until the next restart.
• Adds 5-15s to startup wall-clock when retries back off.
• Compounds with auto-restart loops (when gateway is unstable, every restart redeploys).

Current behavior

• Plugin appears to redeploy every account's slash commands unconditionally on startup.
• No config option in openclaw.json to control this — neither agents.defaults nor channels.discord.accounts.<id> accept a deploy-mode flag.

Requested options

1. channels.discord.slashCommandDeploy: "always" | "changed-only" | "disabled" (default changed-only)
2. Compute a hash of locally registered command definitions and skip deploy when the hash matches the last successful deploy (persisted under ~/.openclaw-secretary/.cache/discord-slash-deploy.json keyed by accountId).
3. When 429 hits, respect retry_after and serialize per-account deploys instead of running them concurrently across accounts.

Workaround

• Reduce gateway restart frequency (already done: update.auto.enabled=false, removed orphaned plugins).
• Cannot fully eliminate without core change.

Repro

1. Configure ≥2 Discord accounts in channels.discord.accounts.
2. Restart the gateway (systemctl --user restart openclaw-<profile>-gateway.service).
3. Observe multiple native slash command deploy warning ... 429 lines within first 30s.