Use video-only test clips for media_perftests video benchmarks.

During development I noticed that the clips we were using contained
audio, which would mean changes to audio-only code were impacting these
benchmarks.

This will cause our tests to use a new baseline, but that's OK.

R=rileya@chromium.org

Review URL: https://codereview.chromium.org/543703003

Cr-Commit-Position: refs/heads/master@{#293325}

media/filters/pipeline_integration_perftest.cc

@@ -68,10 +68,7 @@ TEST(PipelineIntegrationPerfTest, AudioPlaybackBenchmark) {
 }
 
 TEST(PipelineIntegrationPerfTest, VP8PlaybackBenchmark) {
-  RunVideoPlaybackBenchmark("bear-640x360.webm",
-                            "clockless_video_playback_vp8");
-  RunVideoPlaybackBenchmark("bear-320x240.webm",
-                            "clockless_video_playback_vp8");
+  RunVideoPlaybackBenchmark("bear_silent.webm", "clockless_video_playback_vp8");
 }
 
 TEST(PipelineIntegrationPerfTest, VP9PlaybackBenchmark) {
@@ -79,13 +76,13 @@ TEST(PipelineIntegrationPerfTest, VP9PlaybackBenchmark) {
 }
 
 TEST(PipelineIntegrationPerfTest, TheoraPlaybackBenchmark) {
-  RunVideoPlaybackBenchmark("bear.ogv", "clockless_video_playback_theora");
+  RunVideoPlaybackBenchmark("bear_silent.ogv",
+                            "clockless_video_playback_theora");
 }
 
 #if defined(USE_PROPRIETARY_CODECS)
 TEST(PipelineIntegrationPerfTest, MP4PlaybackBenchmark) {
-  RunVideoPlaybackBenchmark("bear-1280x720.mp4",
-                            "clockless_video_playback_mp4");
+  RunVideoPlaybackBenchmark("bear_silent.mp4", "clockless_video_playback_mp4");
 }
 #endif