integration: retry transient docker network ops

Libnetwork endpoint cleanup is eventually consistent. A back-to-back
disconnect+connect on the same network can race teardown and return a
transient error. Wrap the daemon calls in bounded exponential backoff
so TestHASubnetRouterFailoverDockerDisconnect no longer flakes on
phase 4c reconnect.

Fixes #3234

integration/dockertestutil/network_test.go

@@ -0,0 +1,56 @@
+package dockertestutil
+
+import (
+	"context"
+	"errors"
+	"sync/atomic"
+	"testing"
+	"time"
+)
+
+var (
+	errTransientEndpointExists = errors.New("endpoint with name foo already exists in network bar")
+	errPermanent               = errors.New("permanent error")
+)
+
+func TestRetryDockerOp_RecoversFromTransient(t *testing.T) {
+	var attempts atomic.Int32
+
+	op := func() error {
+		if attempts.Add(1) < 3 {
+			return errTransientEndpointExists
+		}
+
+		return nil
+	}
+
+	err := retryDockerOp(context.Background(), op)
+	if err != nil {
+		t.Fatalf("retryDockerOp should recover from 2 transient errors, got: %v", err)
+	}
+
+	if got := attempts.Load(); got != 3 {
+		t.Fatalf("expected 3 attempts, got %d", got)
+	}
+}
+
+func TestRetryDockerOp_RespectsContextCancellation(t *testing.T) {
+	op := func() error {
+		return errPermanent
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 200*time.Millisecond)
+	defer cancel()
+
+	start := time.Now()
+	err := retryDockerOp(ctx, op)
+	elapsed := time.Since(start)
+
+	if err == nil {
+		t.Fatal("retryDockerOp should fail when op always errors")
+	}
+
+	if elapsed > 5*time.Second {
+		t.Fatalf("retryDockerOp should honour ctx deadline (~200ms), took %s", elapsed)
+	}
+}