Respect client retry backoff in supervisor reconciliation

[antongulin]

Fixes #1013.

Impact

The supervisor's periodic 30s reconciliation planned ActionConnect for any enabled, unconnected server regardless of failure history, bypassing the managed client's exponential backoff and MaxConnectionRetries give-up. The ConnectAll path in internal/upstream/manager.go already gates on StateError && !ShouldRetry() — the per-server reconcile path did not. A dead or 401-ing remote upstream was re-dialed every 30 seconds forever (~3 HTTP requests per attempt, ~9k requests/day). In our case this tripped a third-party MCP vendor's server alarms twice and got our account paused (full incident detail in #1013).

Changes

• internal/upstream/types/types.go — extracted the backoff formula from StateManager.ShouldRetry into RetryBackoffDuration(retryCount) (behavior unchanged: 1s → 2s → 4s … capped at 5min), and added ConnectionInfo.ShouldAutoReconnect(now): false while the backoff window is open, after GaveUp/MaxConnectionRetries, and for StatePendingAuth (redialing cannot succeed until the user completes OAuth login; each attempt costs real requests against the upstream).
• internal/runtime/supervisor/supervisor.go — computeReconcilePlan consults actual.ConnectionInfo.ShouldAutoReconnect(...) before planning ActionConnect and plans ActionNone (with a debug log) when the client's retry policy says wait.

Manual reconnects, login flows, reconnect_on_use, and config-change ActionReconnect are unaffected — only the automatic periodic redial is gated. Once the backoff window elapses, the next reconcile tick reconnects as before.

Verification

go test ./internal/upstream/... ./internal/runtime/...   # all pass (incl. new tests)
go vet ./internal/upstream/... ./internal/runtime/supervisor/
gofmt -l <changed files>                                  # clean

New tests:

• TestRetryBackoffDuration — schedule values and overflow caps
• TestConnectionInfo_ShouldAutoReconnect — nil/fresh/ready/pending-auth/backoff-open/backoff-elapsed/gave-up cases
• TestSupervisor_Reconcile_RespectsRetryBackoff — reconcile skips a failed server inside its backoff window, a gave-up server, and a PendingAuth server; reconnects after backoff elapses

Log excerpt of the failure mode this prevents (identical cycle every 30s for two weeks):

18:15:52.759 | INFO  | Starting connection attempt          | {"server": "press-ranger", ...}
18:15:53.809 | ERROR | MCP initialize JSON-RPC call failed  | {"error": "transport error: authorization required"}
18:15:58.214 | ERROR | Connection failed                    | {"error": "OAuth authentication required ..."}
18:16:22.758 | INFO  | Starting connection attempt          | ...