monorepo prep: use contextual cache names, add working_directory input

[matt-codecov]

currently we use cache names like cache-requirements. when worker and api are in different repositories, that's fine - their caches will remain separate. but when we move them all into a monorepo, they will collide. this PR includes the image repo argument in the cache name

our workflows all also assume they are operating out of the repository root. https://github.com/codecov/umbrella, however, will probably organize them in a structure like this:

- apps
  - worker
  - codecov-api
- libs
  - shared

so this PR adds a working_directory input and uses it as the default working directory for all run:-type steps. we can run worker tests in umbrella by calling the run-tests.yml workflow with working_directory: apps/worker. some things to note:

• the default is . so no change should be necessary to the current separate repos. this is just helpful for umbrella
• uses:-type steps (like cache actions) will not use the default working directory. for these steps, paths need to be prefixed with ${{ inputs.working_directory }}/

down the line we should absorb these workflows into the monorepo but fine for now

[trent-codecov]

LGTM. You can release it and opt in your branch to bump the version in api/worker. If that CI works then you know your change is legit.

down the line we should absorb these workflows into the monorepo but fine for now

Maybe but we also use these same workflows in a couple other repos including internal ones. Not saying we are shackled to these but its nice to have it all in one place.

[michelletran-codecov]

I think this will hash a requirements.txt. For migrating to uv, I will need it to hash a uv.lock file instead.

[matt-codecov]

noted. i will skip that for the moment because i don't want it to blow up if the lockfile doesn't exist yet