Add dummy-job mode to improve testing

[in03]

It's not currently possible to queue jobs through Resolve Proxy Encoder without Resolve installed, since we get the job metadata through Resolve's API. While this may seem fair considering targeted use cases, it means development is more difficult. Resolve is not a cheap piece of software and its hardware requirements make development machines expensive too. Beyond development platform requirements, it puts a spanner in the works for CI/CD.

To get around this and still accurately test all down-stream code, we can add a dummy-data mode that mimcs Resolve's API response. We can even ship real test files that the dummy-data references to test the Celery/encoding pipeline. This could be a CLI flag, e.g. --use-dummy-jobs or maybe an environment variable, e.g. RproxUseDummyJobs

If we wanted to bring easier/automated Resolve API testing into the fold too, we could potentially host a local GitHub actions runner on a production machine, or there may be a workflow that allows a GitHub action to run within a pre-provisioned virtual machine. I'm not sure if Resolve would allow itself to be installed and interfaced with in a headless environment though. If not, the local GitHub runner would be the best bet.