Found that this is flaky also on Linux.

https://build.chromium.org/p/chromium.chromiumos/buildstatus?builder=linux-chromeos-rel&number=3038

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/4edb29b3035c2fd521916575e8bc4a9b667f2d65

commit 4edb29b3035c2fd521916575e8bc4a9b667f2d65
Author: Kinuko Yasuda <kinuko@chromium.org>
Date: Wed Nov 29 08:09:01 2017

Disable QuicHttpGoAwayFieldsTest.Misc on all platforms

Started to be flaky after around crrev.com/519987.

TBR=estark@chromium.org
NOTRY=true

Bug: 789430
Change-Id: I45f42f6ac64ef33c0d22c9bd58dd1bdeae531833
No-Presubmit: true
No-Tree-Checks: true
Reviewed-on: https://chromium-review.googlesource.com/795281
Commit-Queue: Kinuko Yasuda <kinuko@chromium.org>
Reviewed-by: Kinuko Yasuda <kinuko@chromium.org>
Cr-Commit-Position: refs/heads/master@{#520039}
[modify] https://crrev.com/4edb29b3035c2fd521916575e8bc4a9b667f2d65/net/quic/http/quic_http_structures_test.cc

crrev.com/519987 is a mechanical rename and shouldn't have had functional changes. I see the test was rolled into chromium around the same time, maybe it was flaky to begin with? Bumping over to ckrasic who did the roll.

Indeed this is just a mechanical rename (albeit within a fork), I am bit stumped as to why this would introduce flakiness.

There will be significant changes to this fork as the HTTP over QUIC mapping diverges from HTTP/2 almost entirely, so disabling this test for now is fine by me.