Improve PR benchmark comment readability

[gavlyukovskiy]

As we have more becnhmarks its getting harder to quickly dissect the results of the PR. We'd like to improve that.

One idea would be to use diff against the results, for example:

Benchmark                                                             (characters)  (jsonPath)  (jsonSize)  (keyLength)  (maskedKeyProbability)  (numberOfTargetKeys)   Mode  Cnt           Score        Error   Units
+ BaselineBenchmark.countBytes                                             unicode         N/A         1kb          N/A                     0.1                   N/A  thrpt    4     2606372.497 ±  97833.565   ops/s
- BaselineBenchmark.countBytes                                             unicode         N/A         1kb                                  0.1                        thrpt    4     2602928.830 ± 150097.202   ops/s

or a markdown table.

We could also use https://github.com/benchmark-action/github-action-benchmark to do that