Draft: Implement "stop after success" scheduler policy

Includes a new command to peek at the next scheduled tasks (in read-only mode).

TODO:

• test coverage
• hook this up as default for the relevant visit types in schedule_recurrent.py

assigned to @olasd

Jenkins job DSCH/gitlab-builds #312 failed in 4 min 44 sec.
See Console Output, Blue Ocean and Coverage Report for more details.

added 1 commit

• e57a29ee - grab_next_visits: implement `stop_after_success` policy

Compare with previous version

Jenkins job DSCH/gitlab-builds #313 failed in 4 min 18 sec.
See Console Output, Blue Ocean and Coverage Report for more details.

added 3 commits

• 7855e234 - Add swh scheduler origin peek-next command
• c0236127 - grab_next_visits: support multiple ORDER BY clauses
• 50362be6 - grab_next_visits: implement `stop_after_success` policy

Compare with previous version

Jenkins job DSCH/gitlab-builds #314 succeeded in 4 min 15 sec.
See Console Output, Blue Ocean and Coverage Report for more details.

Won't that risk to loop over failing origins [1]?

[1] as in degenerated cases (wrong hash referenced or some such) hopefully very few.

Not any more than the current policy. The multiple cooldowns still apply, so in practice we won't re-visit something faster than every failed_cooldown (14 days)

oh yeah, neat, thx.

Sounds like a good idea indeed!