sched: logging improvements (#10550)

This enhances our logging in the scheduler.  The initial "waiting for server" log
no longer claims an initial error state (now "not responding" which better reflects
the actual state).  Runners now have slog wiring to report more details about the
runner, including PID.

llm/server.go

@@ -44,6 +44,7 @@ type LlamaServer interface {
 	EstimatedVRAM() uint64 // Total VRAM across all GPUs
 	EstimatedTotal() uint64
 	EstimatedVRAMByGPU(gpuID string) uint64
+	Pid() int
 }
 
 // llmServer is an instance of the llama.cpp server
@@ -520,6 +521,9 @@ func (s *llmServer) getServerStatus(ctx context.Context) (ServerStatus, error) {
 		if errors.Is(err, context.DeadlineExceeded) {
 			return ServerStatusNotResponding, errors.New("server not responding")
 		}
+		if strings.Contains(err.Error(), "connection refused") {
+			return ServerStatusNotResponding, errors.New("connection refused")
+		}
 		return ServerStatusError, fmt.Errorf("health resp: %w", err)
 	}
 	defer resp.Body.Close()
@@ -640,6 +644,13 @@ func (s *llmServer) WaitUntilRunning(ctx context.Context) error {
 	}
 }
 
+func (s *llmServer) Pid() int {
+	if s.cmd != nil && s.cmd.Process != nil {
+		return s.cmd.Process.Pid
+	}
+	return -1
+}
+
 var grammarJSON = `
 root   ::= object
 value  ::= object | array | string | number | ("true" | "false" | "null") ws