Implement Theil-Sen's method for fitting a line to noisy data (used in bandwidth estimation).

webrtc/base/analytics/percentile_filter_unittest.cc

Index: webrtc/base/analytics/percentile_filter_unittest.cc
diff --git a/webrtc/base/analytics/percentile_filter_unittest.cc b/webrtc/base/analytics/percentile_filter_unittest.cc
index 98168fd27be25da60a7e169b9cd0646a3c41d89f..02b7d091cbc2b9b661594374c43461be0f241ddd 100644
--- a/webrtc/base/analytics/percentile_filter_unittest.cc
+++ b/webrtc/base/analytics/percentile_filter_unittest.cc
@@ -84,15 +84,18 @@ TEST(PercentileFilterTest, MedianFilterUnsigned) {
 TEST_P(PercentileFilterTest, EmptyFilter) {
   EXPECT_EQ(0, filter_.GetPercentileValue());
   filter_.Insert(3);
-  filter_.Erase(3);
+  bool success = filter_.Erase(3);
+  EXPECT_TRUE(success);
   EXPECT_EQ(0, filter_.GetPercentileValue());
 }
 
 TEST_P(PercentileFilterTest, EraseNonExistingElement) {
-  filter_.Erase(3);
+  bool success = filter_.Erase(3);
+  EXPECT_FALSE(success);
   EXPECT_EQ(0, filter_.GetPercentileValue());
   filter_.Insert(4);
-  filter_.Erase(3);
+  success = filter_.Erase(3);
+  EXPECT_FALSE(success);
   EXPECT_EQ(4, filter_.GetPercentileValue());
 }