Update rust-layers, removing a work-around in a test

[mrobinson]

This fixes a regression caused by a previous rust-layers update. We can
now remove the work-around in the reftest.

Fixes #2879.

[hoppipolla-critic-bot]

Critic review: https://critic.hoppipolla.co.uk/r/2107

This is an external review system which you may optionally use for the code review of your pull request.

In order to help critic track your changes, please do not make in-place history rewrites (e.g. via git rebase -i or git commit --amend) when updating this pull request.

[zwarich]

Hmm, this breaks on OS X.

[zwarich]

But it works for me locally, both with and without HiDPI.

[larsbergstrom]

@zwarich The OSX builders are in a virtual machine on Travis - could that be related?

[zwarich]

@larsbergstrom I guess it would probably be using the OpenGL SW renderer rather than virtualizing the GPU, so I should try to force the SW renderer somehow and see if that causes a failure.

[zwarich]

@mrobinson I tried reproducing with every SW GL option I could find, but I had no luck. I find some other failures, which are probably just compositor races.

[mrobinson]

I have temporarily re-enabled the work-around just to ensure that travis now passes and that the test is not simply flaky.

[mrobinson]

Okay. OS X is passing so I suspect that there are multiple regressions here, but one is only appearing on the OS X travis-ci builder.

[mrobinson]

I'm going to abandon this PR for now until I can reproduce the issue that the bots are seeing. The checkout hash of rust-layers will progress regardless in some other PR, but we cannot unskip this test yet.