internal/router/scheduler: add serial scheduler, default on this fork

Add a strict one-model-at-a-time scheduler. Requests run in exact
arrival order; at most one runs at a time; switching to a different
model evicts every other running model first so a single model occupies
memory at a time. Unlike fifo it never reorders or batches same-model
requests, and it ignores group/matrix co-residency entirely, making the
single-model guarantee a property of the scheduler rather than the config.

- new Serial scheduler implementing the Scheduler interface
- register "serial" in scheduler.New; default routing.scheduler.use to
  "serial" at config load (fifo still selectable for upstream behavior)
- update config schema, example config, and config defaults tests

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

internal/config/load.go

@@ -358,11 +358,16 @@ func LoadConfigFromReader(r io.Reader) (Config, error) {
 	config.Routing.Router.Settings.Matrix = config.Matrix
 	config.Routing.Router.Settings.Groups = config.Groups
 
+	// This fork defaults to the "serial" scheduler: one model loaded at a time,
+	// requests served in strict arrival order. Set use: fifo for the upstream
+	// throughput-oriented behavior that batches same-model requests.
 	if config.Routing.Scheduler.Use == "" {
-		config.Routing.Scheduler.Use = "fifo"
+		config.Routing.Scheduler.Use = "serial"
 	}
-	if config.Routing.Scheduler.Use != "fifo" {
-		return Config{}, fmt.Errorf("routing.scheduler.use: unknown scheduler %q (valid: fifo)", config.Routing.Scheduler.Use)
+	switch config.Routing.Scheduler.Use {
+	case "fifo", "serial":
+	default:
+		return Config{}, fmt.Errorf("routing.scheduler.use: unknown scheduler %q (valid: fifo, serial)", config.Routing.Scheduler.Use)
 	}
 	for modelID := range config.Routing.Scheduler.Settings.Fifo.Priority {
 		if _, found := config.RealModelName(modelID); !found {