[WIP]: Revamp CI reporting

[apostasie]

This is an exploratory PR, looking into:

• providing downloadable daemon logs (to allow post-debugging of containerd/etc)
• annotate source code on PRs:
  • provide the list of all failures, linked to their corresponding file
  • if a failing test is known to be flaky (eg: there is an issue named "[Flaky] TestXYZ"), inform the user that it may be ignored - if not, warn the user / reviewer that this failure should not be ignored
• on merge, write a complete report of the run as a comment to a specific collecting ticket, including:
  • list of all failing tests
  • list of tests that did not run
  • pie chart of time spent by packages
  • list of top slow tests
  • split of flaky / not-flaky
  • etc?
• on merge, collate all existing reports from the previous time period (up to the day before), and generate an aggregate report (TODO: figure out concurrency - again)

Fix #4365 and #4370

[apostasie]

.