Reproducing flaky tests about rspecq

[agis]

Some ideas that come to mind:

• use a consistent RSpec seed across all jobs of a certain build (right now we randomize seed before each job execution)
• report the build's seed in Sentry, as part of the flaky test event we emit
• keep track of all the jobs a particular worker has executed (probably to a file), so that we can replay jobs in the same order
• make worker able to consume the above file as input
• make worker optionally accept a --seed flag to use as the RSpec seed for the build
• provide instructions on how to reproduce the failure, probably via Sentry (e.g. rspecq --queue foo.txt --seed 23213)