[blinkpy] Limit WPT manifest generation to <= 25 processes on Windows

We are seeing issues on Windows + Python 3 where a high number of
processes passed to multiprocessing.Pool causes the Pool to crash and
manifest generation to hang. (Very similar or identical to
https://bugs.python.org/issue40263).

This is a temporary local patch to Blink's copy of WPT tooling, in order
to land a workaround as quickly as possible. A proper upstream fix in
WPT should be landed soon, so that we do not have to carry this patch
for too long. Care should be taken in the meantime not to override this
fix if/when we roll Blink tooling.

Bug: 1160108
Change-Id: I8eda8d65caa8cd295ae444676f4628832225f55d
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2601399Reviewed-by: Thomas Guilbert <tguilbert@chromium.org>
Reviewed-by: Rakib Hasan <rmhasan@google.com>
Reviewed-by: Dale Curtis <dalecurtis@chromium.org>
Commit-Queue: Rakib Hasan <rmhasan@google.com>
Auto-Submit: Stephen McGruer <smcgruer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#838996}

third_party/blink/tools/blinkpy/third_party/wpt/wpt/tools/manifest/manifest.py

 import io
 import itertools
 import os
+import sys
 from atomicwrites import atomic_write
 from copy import deepcopy
 from multiprocessing import Pool, cpu_count
@@ -224,10 +225,21 @@ class Manifest(object):
             changed = True
 
         if parallel and len(to_update) > 25 and cpu_count() > 1:
+            # On Windows desktops with high cpu count, we are seeing issues
+            # with high CPU counts and multiprocessing. This is a temporary
+            # workaround until we add logic to WPT to select a max number of
+            # processes. See https://crbug.com/1160108
+            #
+            # 25 was experimentally determined to work in
+            # https://crbug.com/1160108#c22
+            processes = cpu_count()
+            if sys.platform == "win32" and processes > 25:
+              processes = 25
+
             # 25 derived experimentally (2020-01) to be approximately
             # the point at which it is quicker to create Pool and
             # parallelize this
-            pool = Pool()
+            pool = Pool(processes)
 
             # chunksize set > 1 when more than 10000 tests, because
             # chunking is a net-gain once we get to very large numbers