ci: add a mergify automation that notifies us of "stalled" PRs

[thomaseizinger]

Description

Sometimes, a PR with a send-it label stalls due to a failing build. It is easy to forget about those. We should add an automation rule that notifies us in PRs with a send-it label and a failing build. Those will require manual intervention.

Perhaps we could even automatically requeue it once (apply a label to avoid an infinite loop?).

Obviously, we should instead fix the flaky test (looking at you webrtc-rs 👀). In the meantime, this could increase our development velocity.

Related: https://github.com/libp2p/rust-libp2p/issues/3659.

Motivation

Requirements

Open questions

Are you planning to do it yourself in a pull request?

Yes / No / Maybe.

[drHuangMHT]

I looked into mergify documentation and it seems that there is no condition that checks for the duration since a label is applied. However we can check against time when the PR enters merge queue. So here is the question: is the PR put in queue immediate after send-it label is applied?

[jxs]

I looked into mergify documentation and it seems that there is no condition that checks for the duration since a label is applied. However we can check against time when the PR enters merge queue. So here is the question: is the PR put in queue immediate after send-it label is applied?

Hi drHuang, no it needs to fullfill the remaining conditions:

• all conversations resolved
• required CI green