pre-submit tests for github PRs

[Philip Race]

Why is it that the vast majority of PRs are recording spurious looking 
failures of github pre-submit tests ?

https://github.com/openjdk/jdk/pulls?q=type%3Apr+is%3Aopen+label%3Arfr

There seems to be so much noise in these that I pay no attention to them 
any more (except for jcheck)
but they seem to cause concern in some contributors who can't tell what 
they might have broken

Even weirder is that we test Linux x86 there ?? When it isn't a 
mainstream platform. Why ?
Especially since it seem to be the most common failure.

-phil.