Create automatic comparison of performance tests

It should be possible to run performance tests on a new change and not only receive its results, but also compare it to the results before the change.

This might be complex depending on the format of the results, so it's more like a bonus task

No files, tests, or entry points are named. Start by locating how performance tests are run and where their results are stored or displayed, then inspect the result formats and identify how a new run could be compared with the pre-change run. Done means the comparison works for the supported result format, with its limitations documented.

Written by the indexing model from the issue text.