Proof of concept for removing ConstantResolver and using Zeitwerk for ConstantDiscovery

[Catsuko]

What are you trying to accomplish?

I am interested in using Zeitwerk features such as collapsing directories however currently Packwerk doesn't support this.

Based on this comment in a related issue, this PR adds an implementation of ConstantDiscovery using Zeitwerk rather than ConstantResolver as a proof of concept.

What approach did you choose and why?

First I refactored to encapsulate ConstantResolver so it is easier for me to play with an alternate ConstantDiscovery implementation.

Then I implemented a Zeitwerk based ConstantDiscovery and copied tests from ConstantDiscovery and the ConstantResolver gem to make sure it worked ok. Private methods for matching constants from the map of paths was taken from ConstantResolver as well as many tests checking edge cases.

Finally I replaced ConstantDiscovery with my version and removed the code related to the old implementation. A lot of test helpers needed to be updated here since now ConstantDiscovery depends on Zeitwerk Loaders 😭

What should reviewers focus on?

• The second commit contains the zeitwerk implementation as a separate class as well as extended test coverage. I think this should be the focus since the third commit just rigs it all up so I could test on my repo.
• Is it okay to depend on Zeitwerk in this way?

Type of Change

• Bugfix
• New feature
• Non-breaking change (a change that doesn't alter functionality - i.e., code refactor, configs, etc.)

Checklist

• I have updated the documentation accordingly.
• I have added tests to cover my changes.
• It is safe to rollback this change.

[Catsuko]

For usage of Zeitwerk::Loader#all_expected_cpaths: https://github.com/fxn/zeitwerk/blob/main/CHANGELOG.md#2614-14-may-2024

[Catsuko]

I have signed the CLA!

[Catsuko]

For an easier to understand idea of the zeitwerk implementation, see this commit.