feat: correx-managed model lifecycle slice 4 — resource telemetry

Vendor-agnostic ResourceProbe (commons): NvidiaResourceProbe reads nvidia-smi
(injectable runner) for whole-device VRAM/util and /proc/<pid>/status for the
managed llama-server RSS, both fail-soft to null; UnavailableProbe for non-GPU
hosts / the static path. DefaultModelManager.currentPid() + LlamaProcess.pid
expose the managed pid. ServerMessage.ResourceStatus (resource.status,
NonEventMessage, all-nullable) pushed every 2.5s on the global stream plus one
in the initial snapshot. Live gauge only — never event-sourced (feeds no core
decision), so invariants #8/#9 hold. Main wires the NVIDIA probe on the managed
path when nvidia-smi is present, else UnavailableProbe.

Tests: csv parsing (single/multi/malformed), gpu+rss combine, fallbacks.

Plan: docs/plans/2026-05-31-model-lifecycle-management.md (slice 4 of 5).

infrastructure/inference/llama_cpp/src/main/kotlin/com/correx/infrastructure/inference/llama/cpp/DefaultModelManager.kt

@@ -129,6 +129,13 @@ class DefaultModelManager(
 
     override fun currentModel(): ModelDescriptor? = currentDescriptor
 
+    /**
+     * The managed `llama-server` OS pid, or null when no model is resident. Read without the mutex
+     * for the live resource gauge — a stale/null read for one probe tick during a swap is acceptable
+     * (telemetry only, feeds no core decision).
+     */
+    fun currentPid(): Long? = currentProcess?.pid
+
     override fun healthCheck(): ProviderHealth {
         val process = currentProcess ?: return ProviderHealth.Unavailable("No model loaded")
         return if (process.isAlive) {

infrastructure/inference/llama_cpp/src/main/kotlin/com/correx/infrastructure/inference/llama/cpp/LlamaProcess.kt

@@ -16,6 +16,10 @@ class LlamaProcess(
     val isAlive: Boolean
         get() = process?.isAlive ?: false
 
+    /** The OS pid of the running process, or null when not started / not alive. */
+    val pid: Long?
+        get() = process?.takeIf { it.isAlive }?.pid()
+
     suspend fun start() = withContext(Dispatchers.IO) {
         if (isAlive) return@withContext