History is full of uninteresting commits

[jaraco]

Yesterday, I was looking at the repo history and found it inscrutable, due mainly to the "update tox constraints" commits. There are hundreds or thousands of those, creating a branchy history, drowning out meaningful history. While searching for the last tag in the history, I scrolled through many of these commits before giving up.

These commits additionally skew the statistics. Consider git fame:

 cheroot reuse-port $ git fame | head -n 10
Processing: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 172/172 [00:03<00:00, 50.84file/s]
Total commits: 7701
Total ctimes: 7566
Total files: 343
Total loc: 20687
| Author                                         |   loc |   coms |   fils |  distribution   |
|:-----------------------------------------------|------:|-------:|-------:|:----------------|
| Sviatoslav Sydorenko                           | 11265 |   1807 |    113 | 54.5/23.5/32.9  |
| github-actions[bot]                            |  3309 |   2012 |     44 | 16.0/26.1/12.8  |
| Jason R. Coombs                                |  2756 |    996 |     40 | 13.3/12.9/11.7  |
| Kai Mueller                                    |   665 |      7 |     21 | 3.2/ 0.1/ 6.1   |

It appears that github-actions[bot] has the most commits in the repository and the second highest loc.

Let's find a way to stem this madness and restore focus of the primary and secondary value of the project (code and tests).

[webknjaz]

This is my effort to make the CI stable at all times. I've been considering a few things to reduce the noise:

1. Squashing those updates into a single commit (similar to what Dependabot would do)
2. Improving automation so that it would only make pin bumps (merges) when they start breaking the CI (it'd merge the last green bump).

WDYT?

[jaraco]

This problem also affects pull requests. I was looking for a pull request I'd closed recently, but since it was an old one, it was buried under the "Bump transitive deps" PRs.

I've been considering a few things to reduce the noise:

WDYT?

Those approaches seem reasonable. I don't fully understand the design here, but I'm happy to let you work through the issues. I'd like to see this project aim to have less than 20% CI noise (80% of PRs and commits should be for meaningful advancement and not just toil).