In the short term I'll enable the tests and add the failures to expectations. Then I can investigate whether they are fixable or not.

I'm guessing we will need something like fuzzy pixel diffs to really resolve this.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/fd4865395d480f3863f274afcee8f57f206c81c0

commit fd4865395d480f3863f274afcee8f57f206c81c0
Author: Stephen Chenney <schenney@chromium.org>
Date: Tue Aug 08 17:48:08 2017

Claim ownership of wpt tests for the paint team

The tests are now part of Blink regular test suite, and the
paint team owns the functionality that is tested. So we should
own the tests.

R=foolip@chromium.org,chrishtr@chromium.org
BUG=753080

Change-Id: Ic6d903c9e9ab4ac1a3abbbd3a1d9ba9e8141c98b
Reviewed-on: https://chromium-review.googlesource.com/606167
Reviewed-by: Philip Jägenstedt <foolip@chromium.org>
Reviewed-by: Chris Harrelson <chrishtr@chromium.org>
Commit-Queue: Stephen Chenney <schenney@chromium.org>
Cr-Commit-Position: refs/heads/master@{#492693}
[add] https://crrev.com/fd4865395d480f3863f274afcee8f57f206c81c0/third_party/WebKit/LayoutTests/external/wpt/css/css-transforms-2/OWNERS
[add] https://crrev.com/fd4865395d480f3863f274afcee8f57f206c81c0/third_party/WebKit/LayoutTests/external/wpt/images/OWNERS
[modify] https://crrev.com/fd4865395d480f3863f274afcee8f57f206c81c0/third_party/WebKit/LayoutTests/external/wpt/svg/OWNERS

My informal breakdown would be as follows:
Expectations differ from Blink's rendering by 1px.
external/wpt/css/css-transforms-2/backface-visibility-hidden-001.html
external/wpt/css/css-transforms-2/rotate_x_45deg.html
external/wpt/css/css-transforms-2/rotate_y_45deg.html
external/wpt/css/css-transforms-2/transform3d-preserve3d-010.html
external/wpt/css/css-transforms-2/transform3d-preserve3d-013.html
external/wpt/css/css-transforms-2/transform3d-scale-005.html
external/wpt/css/css-transforms-2/transform3d-scale-006.html
external/wpt/css/css-transforms-2/transform-3d-rotateY-stair-below-001.xht

No transformed image in expectation
external/wpt/css/css-transforms-2/transform3d-image-scale-001.html
external/wpt/css/css-transforms-2/transform3d-image-scale-002.html

Expectation appears to lack perspective
external/wpt/css/css-transforms-2/transform3d-perspective-003.html
external/wpt/css/css-transforms-2/transform3d-perspective-004.html
external/wpt/css/css-transforms-2/transform3d-perspective-005.html

Different sorting order
external/wpt/css/css-transforms-2/transform3d-sorting-002.html
external/wpt/css/css-transforms-2/transform3d-sorting-004.html
external/wpt/css/css-transforms-2/transform3d-sorting-006.html
external/wpt/css/css-transforms-2/ttwf-css-3d-polygon-cycle-mismatch.html
external/wpt/css/css-transforms-2/ttwf-css-3d-polygon-cycle.html

Other mismatch
external/wpt/css/css-transforms-2/transform3d-matrix3d-001.html
external/wpt/css/css-transforms-2/transform3d-rotatex-perspective-003.html
external/wpt/css/css-transforms-2/transform3d-scale-007.html

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c5a1bfee03cbc993159d08b2d4212dd76519a9ad

commit c5a1bfee03cbc993159d08b2d4212dd76519a9ad
Author: Eric Willigers <ericwilligers@chromium.org>
Date: Fri Aug 25 21:51:55 2017

css-transforms-2: Enable most web platform tests

Instead of disabling all the tests, we now disable only
the tests that are known to fail.

Preliminary analysis by Stephen Chenney:
https://groups.google.com/a/chromium.org/d/msg/blink-dev/bY4v78aN8to/eLUPbq__AQAJ

BUG=753080

Change-Id: I162c7bdfa9fbe44f13d87fea719cd660d113a311
Reviewed-on: https://chromium-review.googlesource.com/634783
Reviewed-by: Stephen Chenney <schenney@chromium.org>
Commit-Queue: Eric Willigers <ericwilligers@chromium.org>
Cr-Commit-Position: refs/heads/master@{#497552}
[modify] https://crrev.com/c5a1bfee03cbc993159d08b2d4212dd76519a9ad/third_party/WebKit/LayoutTests/NeverFixTests

For tests like transform3d-scale-005.html, should we have exact rotations by multiples of 90degrees?

We could identify such cases by computing mod with 90 before converting to radians.

Re comment 5: is transform3d-scale-005.html passing on Firefox? I'm pretty
sure that Firefox snaps transforms to pixels, whereas Chrome snaps pre-transform
and then applies the transform as a floating-point mapping. This is a real
and significant compatibility issue.