Guided remote cache flow

[adhorodyski]

Describe the feature

Our current remote cache flow (rock run:*) requires manual debugging steps in order to understand a cache miss when it occurs.
This is not only complex, but assumes a user has high understanding of all the parts involved in this process, including debugging with CI logs on the most recent artifacts and comparing fingerprint outputs (not to say being able to manually test them in the first place).

I propose a more guided workflow that helps you not only understand the whole process (step by step), but allows you to easily recover or even recovers for you whenever possible. This can include:

• listing recent (and actively running) CI jobs with their fingerprints
• some variation of a fingerprint --diff command to only highlight the files changed
• any form of git clean ran interactively when it's possible with no harm
• make the above commands an interactive part of anyrun:* command for fast tracked recovery

Motivation

Provide users with the best DX possible.

Related Issues

[thymikee]

Let's think about it:

• listing workflows would require the CLI to know which CI you're running on (we're currently CI agnostic, the CLI only has information about the remote cache provider) and integrating with that
• fingerprint diffing could be done by keeping build metadata (e.g. fingerprint sources) next to the build: in the artifact zip or/and as a separate entry (so that we can quickly retrieve lightweight metadata without downloading the whole build)
• it will happen that there really isn't a build that matches your local state
• because of the above I wouldn't like to integrate it as a part of run:* commands to avoid extra clutter (unless behind a flag). I'd see it rather as remote-cache debug or fingerprint --diff or so.

[adhorodyski]

Discussed in Slack, here's a refined version of this workflow that @thymikee suggested:

1. cache miss
2. if there ever was a remote cache hit, ask whether they want to debug this or continue with local build?
3. create copy of ios/android folder in os.tmpdir()
4. run git clean -fdx
5. run remote-cache list again to see if the artifact is available
6. inform user that we've verified that the build is available
7. prompt: can we call git clean -fdx for you and retry the run:ios/run:android?

[adhorodyski]

I think I was able to vibe code this one over the weekend, will test within a bigger codebase and update with a PR once ready.