Add support for specifying artifacts via `artifact-id`

[julealgon]

Describe

When specifying an artifact for download, the current implementation only supports passing the artifact's name. In some situations, this creates unnecessary code duplication as one needs to provide the same name as when the upload happened in another job.

If the artifact name then needs to be changed later, it needs to be updated in both places, making it fragile.

Proposed solution

I propose that the action also takes an artifact-id and use that for downloading the artifact, on top of the existing name-based API.

Alternatives considered

At the moment, we are hardcoding the name on the upload step and in the report step. An alternative to that would be to explicitly use step and job outputs to create the artifact name in a single place, then use step output in the upload, and job output in the report.

However, this introduces a step just to expose the artifact name, which is more convoluted than just taking the artifact-id as output directly from the upload job and propagate that.

Another alternative is to not rely on this action's download capabilities and have an explicit download step, passing the artifact-id to that.

[jozefizso]

How would artifact-id solve this problem?

I personally prefer to have a step which computes the variables and they are used across other steps.

[julealgon]

@jozefizso , the upload-artifact action provides the created artifact-id as one of its outputs, so it can more easily be integrated with other steps/jobs without requiring an extra step to explicitly expose the variables as environment variables and/or step outputs.

What you described is one of the possible workarounds I mentioned in the OP.