🐛 Prevent race when informers are started more than once

If `Informers` are started a second time, there is a possibility for a
data race because it sets a `ctx` field on itself. This write is
protected by a mutex, but reads from that field are not.

pkg/cache/internal/informers.go

@@ -18,6 +18,7 @@ package internal
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"math/rand"
 	"net/http"
@@ -186,6 +187,12 @@ type Informers struct {
 // Start calls Run on each of the informers and sets started to true. Blocks on the context.
 // It doesn't return start because it can't return an error, and it's not a runnable directly.
 func (ip *Informers) Start(ctx context.Context) error {
+	select {
+	case <-ip.startWait:
+		return errors.New("Informrmer already started")
+	default:
+		// do nothing
+	}
 	func() {
 		ip.mu.Lock()
 		defer ip.mu.Unlock()