fix: ensure validation is run on correct files

[beckermr]

This PR ensures we run output validation only on the packages we built.

closes #342
closes #341

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

[beckermr]

cc @isuruf

[beckermr]

ok @isuruf I've added caching and I added the options to the CI setup calls. They are not required right now. Once we adjust smithy, we can turn them to be required if we like.

[jaimergp]

I think this broke inspect_artifacts in staged-recipes. Not sure why we need to render the recipes there (again). See here for Linux logs (fails but error is ignored):

Inspecting artifacts
+ command -v inspect_artifacts
+ inspect_artifacts
Error: non-existent: /home/conda/staged-recipes/recipe
+ echo 'inspect_artifacts needs conda-forge-ci-setup >=4.6.0'
inspect_artifacts needs conda-forge-ci-setup >=4.6.0

The Error: non-existent string comes from conda_build.render. I guess in staged-recipes we can't pass /recipe blindly, since the recipe dir depends on the PR contents.

[beckermr]

Ah I see. The goal in this pr was to not print inspection for every artifact but only the ones that were new. When there are downstream tests, we inspect the downstreams too which is odd and confusing.

[jaimergp]

Got it. I think we raaaarely use downstreams in staged-recipes, right? Should we special case it? Or adjust this so it accepts several directories and pass recipes/* from staged-recipes?