[server] Add coordinator metrics for rebalance

### Search before asking

- [x] I searched in the [issues](https://github.com/apache/fluss/issues) and found nothing similar.

### Description

### Motivation

Rebalance currently exposes no metrics at all. The only way to observe a rebalance is to poll `Admin#listRebalanceProgress` or tail coordinator logs. #3965 (PR #3985) improves the Admin API with timestamps and historical listing, but that covers on-demand queries, not continuous monitoring. Operators cannot:

- Alert when a rebalance is running, stuck, or failing.
- Chart migration throughput and per-bucket duration on a dashboard.
- Correlate a rebalance window with write-latency or disk-usage changes on TabletServers.

The coordinator already registers cluster gauges such as `bucketCount` through `CoordinatorMetricGroup` (`fluss-server/.../metrics/group/CoordinatorMetricGroup.java`), and `RebalanceManager` already tracks all the needed state (`inProgressRebalanceTasksQueue`, `inProgressRebalanceTasks`, `finishedRebalanceTasks`, the in-flight task start timestamp), so this is a wiring task, not a new subsystem.

### Proposed solution

Register a rebalance metric set on the coordinator metric group, with names added to `MetricNames`:

Gauges for the current task:

- `rebalanceInProgress` (0 or 1)
- `rebalanceBucketsPending`
- `rebalanceBucketsCompleted`, `rebalanceBucketsFailed`, `rebalanceBucketsTimedOut`
(split of the finished map by per-bucket status)
- `rebalanceDurationMs` (time since the task was registered, 0 when idle)
- `inflightBucketDurationMs` (time the current bucket task has been running, 0 when idle)

Counters across tasks (survive task completion, reset on coordinator failover):

- `rebalancesCompletedTotal`, `rebalancesFailedTotal`, `rebalancesCanceledTotal`

Implementation notes:

- Expose the values from `RebalanceManager` via supplier gauges. The backing maps are `ConcurrentHashMap` and the in-flight fields are volatile, so reads from the metrics reporter thread are safe without touching the coordinator event thread.
- Register on `CoordinatorMetricGroup`, following the existing `bucketCount` pattern.
- Document the new metrics in `website/docs/maintenance/observability/monitor-metrics.md`.

Out of scope: per-table or per-bucket metric groups (cardinality), and the Admin API side already covered by #3965.

### Willingness to contribute

- [x] I'm willing to submit a PR!

Start with fluss-server/.../metrics/group/CoordinatorMetricGroup.java and its existing bucketCount registration, then inspect RebalanceManager and MetricNames for the tracked state and metric naming. Document the metrics in website/docs/maintenance/observability/monitor-metrics.md. Done means the coordinator exposes the requested gauges and counters with the stated idle, status, and failover behavior.

Written by the indexing model from the issue text.