Add short Node Monitoring page to website (#223)

Author: dgrove-oss

site/_data/navigation.yml

@@ -32,3 +32,5 @@ side:
     url: /arch-controller/
   - title: Fault Tolerance
     url: /arch-fault-tolerance/
+  - title: Node Monitoring
+    url: /arch-node-monitoring/

site/_pages/arch-node-monitoring.md

@@ -0,0 +1,28 @@
+---
+permalink: /arch-node-monitoring/
+title: "Node Monitoring"
+classes: wide
+---
+
+The AppWrapper controller can optionally monitor Kubernetes Nodes and
+dynamically adjust the `lendingLimits` on a designated `ClusterQueue`
+to account for dynamically unavailable resources. This capability is
+designed to enable cluster admins of an
+[MLBatch cluster](https://github.com/project-codeflare/mlbatch) to fully
+automate the small scale quota adjustments required to maintain full cluster
+utilization in the presence of isolated node failures and/or
+minor maintenance activities.  The monitoring detects both Nodes that
+are marked as `Unscheduable` via standard Kubernetes mechanisms and Nodes
+that have resources that Autopilot has flagged as unhealthy (see [Fault Tolerance](/arch-fault-tolerance)).
+The `lendingLimit` of a designated slack capacity `ClusterQueue` is
+automatically adjusted to reflect the current dynamically unavailable resources.
+
+Node monitoring is enabled by the following additional configuration:
+```yaml
+slackQueueName: "slack-queue"
+autopilot:
+  monitorNodes: true
+```
+
+See [node_health_monitor.go]({{ site.gh_main_url }}/internal/controller/appwrapper/node_health_monitor.go)
+for the implementation.

site/_pages/dev-setup.md

@@ -6,7 +6,7 @@ classes: wide
 
 ### Prerequisites
 
-You'll need `go` v1.22.0+ installed on your development machine.
+You'll need `go` v1.22.4+ installed on your development machine.
 
 You'll need a container runtime and cli (eg `docker` or `rancher-desktop`).