Add dev release using a reusable workflow

[MonicaOlejniczak]

↪️ Pull Request

These changes introduce a new dev release workflow that can be manually triggered in Github Actions. It aims to give contributors a way to validate changes not yet on the main branch.

The dev-release workflow calls a reusable release workflow that performs the majority of the work, and will soon begin to replace the tag-release and nightly-release workflows in the same way. Other additions include:

• Merge build-apple-silicon into the build job (now renamed to build-macos-windows)
• Remove JEMALLOC_SYS_WITH_LG_PAGE that existed in build-apple-silicon as the underlying issue should be fixed here
• Add Rust caching to non-dockerized jobs
  • aarch64-apple-darwin
  • macos-latest and windows-latest (this will match the CI cache names, though I did not validate whether they'll get used or if it breaches Githubs cache isolation)
  • arm-unknown-linux-gnueabihf
  • aarch64-unknown-linux-gnu

💻 Examples

Observe the successful workflow build here

🚨 Test instructions

Test on your own branch by adding a push trigger with your branch name to dev-release, as you cannot manually trigger an workflow until it is on the main branch

[mattcompiles]

Awesome work 👏 🍻

[marcins]

Add Rust caching to non-dockerized jobs

Which jobs are dockerized, and is it just a limitation of GH Actions that you can't do the caching for those jobs? Does that have any material impact on timings?

[marcins]

Nice work!

[MonicaOlejniczak]

@marcins I read that caching is not very effective for docker containers in one of the actions documentation, but did not test it out to confirm. I can verify the behaviour as a follow up, but also am wary of blowing up the small cache size in Github.

[devongovett]

FYI I think you still need JEMALLOC_SYS_WITH_LG_PAGE. If you try to require the built binary on an Apple Silicon Mac you'll get the following error:

> require('./parcel-node-bindings.darwin-arm64.node')
<jemalloc>: Unsupported system page size
memory allocation of 216 bytes failed
zsh: abort      node

[MonicaOlejniczak]

@devongovett Okay, guess I'll find out soon enough! It's easy enough to add back in