The tests:

paint/invalidation/multicol-with-abspos.html
paint/invalidation/selection-change-in-iframe-with-relative-parent.html
paint/invalidation/multicol-with-inline.html
paint/invalidation/textarea-set-disabled.html
paint/invalidation/multicol-with-abspos-in-relpos.html
paint/invalidation/table-hover-on-link.html
paint/invalidation/multicol-as-paint-container.html
paint/invalidation/paint-caret-in-div-with-negative-indent.html
paint/invalidation/multicol-with-text.html
paint/invalidation/multicol-with-block.html
paint/invalidation/inline-outline-repaint-2.html
paint/invalidation/inline-reflow.html
paint/invalidation/multicol-repaint.html

appear to be flaky when run in random order. Steps to reproduce:

This should pass consistently:
Tools/Scripts/run-layout-tests $TEST_DIRECTORY

This should fail intermittently:
Tools/Scripts/run-layout-tests $TEST_DIRECTORY --order=random

They failed in 11 out of 63 runs on the WebKit Win - RandomOrder builder. Here's an example of a build where they failed:
https://luci-milo.appspot.com/buildbot/chromium.fyi/WebKit%20Win%20-%20RandomOrder/187

We're working on removing order dependency from Blink layout tests. More context here:
https://docs.google.com/document/d/11-628Ny6hY__kfIv__zVD0m3rA0OB9tQLLC_gJI_PNg/edit