Consider caching for job trace summaries
What
The new pipeline Job Failures tab includes summaries of failed build traces. Fragment caching could help with this, as mentioned in https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/10719#note_28893634
Why
Performance. Each job trace will currently be read from the end of a file on whichever drive or network location build summaries are held in.
We might want to monitor actual production timing to see how fast/slow it currently is first though.
At the moment we limit the page to 10 summaries for performance. Caching these might allow us to increase that limit too.