[client_lb_e2e_test] increase timeout for WeightedRoundRobinTest.WithOutlierDetection test (#37718)

This attempts to fix the following flake:

https://btx.cloud.google.com/invocations/e8a6ff31-ba5f-49ff-97d7-eb4b6b3b7c04/targets/%2F%2Ftest%2Fcpp%2Fend2end:client_lb_end2end_test@poller%3Dpoll;config=c4ae5af353698403518bd66f686ce4f7f10d865e4cdcccbb7036582cbc9fa7d6/tests

The flake is a timeout, and it's extremely rare -- it's happened only twice in the last year, and both times under MSAN.  I can't reproduce it manually, even running 1000x on RBE with MSAN.  But I think this PR should address it.

Closes #37718

COPYBARA_INTEGRATE_REVIEW=https://github.com/grpc/grpc/pull/37718 from markdroth:client_lb_e2e_flake 1ee75c7767
PiperOrigin-RevId: 674387421

test/cpp/end2end/client_lb_end2end_test.cc

@@ -3277,7 +3277,7 @@ TEST_F(WeightedRoundRobinTest, CallAndServerMetric) {
 // all of its subchannels every time it saw an update, thus causing the
 // WRR policy to re-enter the blackout period for that address.
 TEST_F(WeightedRoundRobinTest, WithOutlierDetection) {
-  const int kBlackoutPeriodSeconds = 5;
+  const int kBlackoutPeriodSeconds = 10;
   const int kNumServers = 3;
   StartServers(kNumServers);
   // Report server metrics that should give 6:4:3 WRR picks.