Parallelize fetching telemetry benchmark deps

The fetching script currently fetches about 2 GB (on my linux workstation)
from cloud storage serially. This can (and is) trivially parallelizable, and
should speed up the fetching script significantly.

This script has been taking 20 minutes on mac machines for a long time. This
will hopefully significantly speed the mac builders up.

Bug: 1102597
Change-Id: I93d2fa2b1f4b4ef6e27e9c894a9dd4c0d0e12d7b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2285503
Commit-Queue: Stephen Martinis <martiniss@chromium.org>
Reviewed-by: Wenbin Zhang <wenbinzhang@google.com>
Reviewed-by: John Chen <johnchen@chromium.org>
Cr-Commit-Position: refs/heads/master@{#786521}

tools/perf/fetch_benchmark_deps.py

@@ -7,10 +7,11 @@
 
 import argparse
 import json
+import logging
+import multiprocessing
 import optparse
 import os
 import sys
-import logging
 
 from chrome_telemetry_build import chromium_config
 from core import benchmark_finders
@@ -121,8 +122,12 @@ def main(args):
       raw_input(
           'No benchmark name is specified. Fetching all benchmark deps. '
           'Press enter to continue...')
-    for b in benchmark_finders.GetOfficialBenchmarks():
-      deps[b.Name()] = _FetchDepsForBenchmark(b)
+    benchmarks = benchmark_finders.GetOfficialBenchmarks()
+    p = multiprocessing.Pool()
+    results = p.map(_FetchDepsForBenchmark, benchmarks)
+    p.terminate()
+    for benchmark, result in zip(benchmarks, results):
+      deps[benchmark.Name()] = result
 
   if options.output_deps:
     with open(options.output_deps, 'w') as outfile: