Fix parsing of HTML files

HTML files collected by running systrace can have multiple trace-data
section in them. Currently we can only parse the section with raw
systrace output in it. This CL allows us to skip the other sections.

Also fixed a bug in systrace parsing causing pids not to be read
correctly.

Bug: 143202192
Change-Id: Ia2f88cf1452c1d46c870a73c540bfe0210b95759
diff --git a/tools/install-build-deps b/tools/install-build-deps
index 5cd6e31..26a2e1a 100755
--- a/tools/install-build-deps
+++ b/tools/install-build-deps
@@ -145,8 +145,8 @@
     # Example traces for regression tests.
     (
         'buildtools/test_data.zip',
-        'https://storage.googleapis.com/perfetto/test-data-20191017-212419.zip',
-        '7c0efa891fe85202c59391327538b327daa1a0f5',
+        'https://storage.googleapis.com/perfetto/test-data-20191025-102200.zip',
+        '25a61211f82cd73c6ce5b67da9d73f046de7b3f9',
         'all',
     ),