Add Workflow to check if we break snmalloc-rs

[mjp41]

Currently the downstream consumer snmalloc-rs is very dependent on the internals of snmalloc. This strong coupling makes release a little awkward. We either need

• Release process for snmalloc, should have a PR ready to go on snmalloc-rs before we hit complete.
• Add some workflow jobs that replicate snmalloc-rs CI, and check each PR doesn't break it. (Allow failure, but need to track)

[SchrodingerZhu]

If needed, as the owner of snmalloc-rs, I am happy to donate the whole repo to upstream and I am willing to do the most work to merge it into microsoft/snmalloc.

The caveat is that we will lose the native FreeBSD CI for rust.

[SchrodingerZhu]

A PoC in made in #744.

[mjp41]

Thanks for doing this. Are you sure about this. You've put a lot of work into snmalloc-rs, and putting it under the Microsoft org, would give you less personal ownership.

I think we should think carefully about the pros and cons of merging it in. I'll try to get the CI functioning, but I propose we leave it a week or two to think about it.

[SchrodingerZhu]

but I propose we leave it a week or two to think about it.

Good idea. I would also like to collect others' idea

You've put a lot of work into snmalloc-rs, and putting it under the Microsoft org, would give you less personal ownership.

Personally, I think it is okay as long as doing this reduces "unsync" between snmalloc and snmalloc-rs. There is a large amount of downloads of snmalloc-rs so I think to improve its maintainability is of great importance now.

Anyway, let's take time to think about the best way to handle this. Also, another thing I want to do is to sync the version and release cycle between upstream and downstream.