DRY out the runner lifecycle code (#12540)

* DRY out the runner lifecycle code

Now that discovery uses the runners as well, this unifies the runner spawning code
into a single place.  This also unifies GPU discovery types with the newer ml.DeviceInfo

* win: make incremental builds better

Place build artifacts in discrete directories so incremental builds don't have to start fresh

* Adjust sort order to consider iGPUs

* handle cpu inference oom scenarios

* review comments

discover/cpu_linux_test.go

@@ -2065,12 +2065,6 @@ power management:
 			cpus := linuxCPUDetails(buf)
 
 			slog.Info("example", "scenario", k, "cpus", cpus)
-			si := SystemInfo{
-				System: CPUInfo{
-					CPUs: cpus,
-				},
-			}
-			threadCount := si.GetOptimalThreadCount()
 			if len(v.expCPUs) != len(cpus) {
 				t.Fatalf("incorrect number of sockets: expected:%v got:%v", v.expCPUs, cpus)
 			}
@@ -2085,10 +2079,6 @@ power management:
 					t.Fatalf("incorrect number of threads: expected:%v got:%v", v.expCPUs[i], c)
 				}
 			}
-
-			if threadCount != v.expThreadCount {
-				t.Fatalf("incorrect thread count expected:%d got:%d", v.expThreadCount, threadCount)
-			}
 		})
 	}
 }