Add one more benchmark plot on recording for cache only

[lorenzwalthert]

This PR

• switches to checkpoint of RStudio Package Manager and hence fix all versions of all dependencies for better comparison.
• Wanted to introduce a PR comment bot but it did not work with various action providers because apparently, commenting on PRs from forks is not really solved with GitHub Actions yet.
• introduced a new benchmark for first stying with empty cache, tied to a git reset HEAD --hard, called cache_recording. It reflects the situation when you have the cache feature turned on and style code that was never styled before. This is the most common case I think.
• Make the action where the cache is activated (cache_recording) comparable with where it's not (without_cache) by simulate the git reset in the latter (so they are equal up to the caching feature).

[codecov-io]

Codecov Report

Merging #684 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #684   +/-   ##
=======================================
  Coverage   90.43%   90.43%           
=======================================
  Files          47       47           
  Lines        2226     2226           
=======================================
  Hits         2013     2013           
  Misses        213      213           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c50338f...5f34131. Read the comment docs.