Make it easier to identify why the grind test might be failing and increase timeout for truffleruby

[rwstauner]

This adds some additional checks and messages to help identify why the test is failing and what might help.

When I first ran the suite with TruffleRuby I was seeing

This worker is exiting early because it reached its timeout of 1 seconds
  test_grind_max_time                                            ERROR (8.89s)
Minitest::UnexpectedError:         NoMethodError: undefined method `scan' for nil:NilClass
            /home/spin/src/github.com/Shopify/ci-queue/ruby/test/integration/grind_redis_test.rb:131:in `test_grind_max_time'

This PR:

• captures that STDERR line
• asserts that it is present (to ensure that the subprocess exited for the timeout as expected)
• protects against nil error with a few more assertions
• adds a message about what might be happening and what might help
• measures the time the process took for when its useful to know (based on ruby versions i can get a range including ~2s, ~7s, ~17s)
• increases the timeout for when using truffleruby to something that will allow the subprocess to get started so that the tests pass

[casperisfine]

I don't see a big value in this.

Like o that was your issue with Truffle, but what is the chance someone else run into the exact same issue in the future?

[rwstauner]

I've run into enough tests that fail or error without providing any hints as to what might be wrong
that when i come upon them I like to make them clearer for the next person.
If there was a reason that it didn't pass this time, then it could probably happen again.

It was a while back but I think I originally went down a longer (and less fruitful) road trying to understand what might be wrong (when I thought there might be some other reason that the subprocess wasn't working) before I tried simply increasing the timeout.

[rwstauner]

just got this on a branch in CI which was very helpful to see