Consider warning X amount of time after push but PR is still marked as "waiting on author"

from comment down below:

Recently, triagebot started responding to @rustbot author with a comment that reminds the comment sender to use @rustbot ready in the future. But it was deemed to be too spammy. So now, after https://github.com/rust-lang/triagebot/pull/1964, the response is only sent to users that are not members of the given repository.

This issue essentially proposes to send the response to everyone, but only once in e.g. 30 days or something, as a periodic reminder that @rustbot ready exists and should be used.

Original discussion started here, https://github.com/rust-lang/triagebot/pull/1964#issuecomment-2851753994:

@Kobzol:

2. Actually detect the situations where a PR's last action is commit by the user, but it is still waiting on author. E.g., when a PR is S-waiting-on-author, it received a commit/push after the latest state change, and >2 weeks have received, try to ping the PR author about @bot ready.

@RalfJung:

No strong opinion either way... but regarding (2), I think 2 weeks is way too much. I'd make it more like 24h.

@Urgau:

Yeah, your second option is ideally what I would want. Also agree with Ralf, that 2 weeks is too long, 24 hrs seems fine. We should also make sure CI is green and other stuffs, like S-blocked.

@Kobzol:

I used 2 weeks as an example, because I often push something that is WIP, but don't want to mark it as ready for review yet, as I'm planning to push more stuff before re-requesting review. So 24h would be "too fast". It's in general difficult to recognizehttps://github.com/rust-lang/triagebot/issues/1978#issuecomment-2866358129 if a PR is really ready for a review :)

@RalfJung:

I think most contributors would be very confused and somewhat frustrated if they push and then nothing happens for 2 weeks and then they get told what they should have done 2 weeks ago. So for the typical sporadic contributor, 2 weeks is way too much IMO -- even 24h seems long.

@Urgau:

I used 2 weeks as an example, because I often push something that is WIP, but don't want to mark it as ready for review yet, as I'm planning to push more stuff before re-requesting review. So 24h would be "too fast".

Maybe your workflow would be better represented by using the PR draft state? We could simply not issue the warning if the PR is in draft.

@Kobzol:

If we take that to the extreme, we could just use draft for "waiting on author", that would make the "I'm waiting for author" situation very obvious and visual :)

I think that draft PRs shouldn't even be in "S-waiting-on-review" often now that we don't auto assign draft PRs.

Start by reviewing the behavior and discussion in pull request 1964, especially how triagebot handles @rustbot author, @rustbot ready, and the S-waiting-on-author state. The issue names no files or tests, and the warning timing and conditions need agreement before implementation; done would mean an agreed, tested reminder policy.