docs: five exposed metrics are missing from the monitoring reference

## Problem

The controller registers 16 Prometheus metrics, but the monitoring reference at `docs/book/src/operations/monitoring.md` only documents 11 of them. Five metrics are served on `/metrics` with no matching entry in the docs, so operators have no name, type, or label reference for them.

## Undocumented metrics

All five are registered in the `MustRegister` block at `internal/metrics/metrics.go:186-189`:

| Metric | Type | Labels | Definition |
| --- | --- | --- | --- |
| `node_readiness_bootstrap_duration_seconds` | Histogram | `rule` | `internal/metrics/metrics.go:116` |
| `node_readiness_reconciliation_latency_seconds` | Histogram | `rule`, `operation` | `internal/metrics/metrics.go:126` |
| `node_readiness_nodes_by_state` | Gauge | `rule`, `state` | `internal/metrics/metrics.go:138` |
| `node_readiness_condition_failures_total` | Counter | `rule`, `condition` | `internal/metrics/metrics.go:148` |
| `node_readiness_rule_last_reconciliation_timestamp_seconds` | Gauge | `rule` | `internal/metrics/metrics.go:158` |

## Fix

Add each metric to `docs/book/src/operations/monitoring.md` using the format the existing entries follow: name, type, labels, and a one-line description of what it measures. The `Help` string and label list in the source give the wording.

## Done when

- All 16 registered `node_readiness_*` metrics appear in `monitoring.md`.
- Each entry lists the type and labels that match `internal/metrics/metrics.go`.

Start with the existing metric entries in docs/book/src/operations/monitoring.md, then compare the Help strings, types, and labels in internal/metrics/metrics.go at the listed definitions and MustRegister block. Add the five missing metrics using the existing format; done means all 16 registered node_readiness_* metrics appear with matching types, labels, and descriptions.

Written by the indexing model from the issue text.