Use the term 'relevant' instead of 'significant' in try run results

[rylev]

Fixes #968

This changes the try run result in two ways:

• Uses the term "relevant" instead of "significant". Relevant better captures what the results are: results that we should care about in someway. "Significant" is a bit too overloaded. It can also mean "large" and "statistically significant" which is not the criteria we use for when to show changes in try run results.
• It also adds the magnitude of the largest change to the message to give the reader an idea of how severe the changes are.

The changes in #968 would produce the following message under these changes:
"This change led to small relevant regressions 😿 in compiler performance."

[Mark-Simulacrum]

"This change led to small relevant regressions crying_cat_face in compiler performance."

It might be worth a link or some connector here in the future -- e.g., "This change led to small, but still relevant, regressions..." I think relevant is also "interesting" in that there's an open question of 'relevant to whom?', in some sense. I think having a paragraph for the current rationale linked from the relevant word that talks about our method for determining what is relevant and how precisely authors/reviewers should judge those results would be good. For example, it is likely useful to discuss there that small wall time changes are still relevant and shouldn't be dismissed out of hand simply due to their low magnitude (but rather, for example, lack of impact on non-stress-test benchmarks, or some other reasoning).