feat: multi arch and file-based OLM catalog

[raelga]

• Updates all build targets with support for arm64 and amd64 images. 6b36cf9
• Adds the catalog-* targets to generate file-based OLM catalogs. 6b36cf9
• Adds the release-catalog GitHub workflow, to generate and build a catalog for the operator. 6b36cf9
• Generates the complete catalog with all bundle releases from v0.3.8 to v0.10.7. cf94821
• Updates the release documentation with the new targets. abbdada
• Removes the build step in the test workflow a3ac200

Related to:

• ARM64 images and bundling for external-secrets-helm-operator external-secrets#3930
• https://olm.operatorframework.io/docs/reference/file-based-catalogs/

[Skarlso]

Hello. :) You have a test failure:

Error: must build at directory: not a valid directory: evalsymlink failure on 'config/testing' : lstat /home/runner/work/external-secrets-helm-operator/external-secrets-helm-operator/config/testing: no such file or directory
error: no objects passed to apply

[raelga]

@Skarlso It should be working now

[gusfcarvalho]

Hi @raelga . We internally have been discussing the feasibility of maintaining OLM releases within our team for a while now, and it is clear from past releases (external-secrets/external-secrets#4036) we are not par to maintaining and further releasing OLM builds anymore.

On the community meeting today I have proposed external-secrets/external-secrets#4175 and it was approved.

While this does not mean we are not having any support for OLM in the future, it is clear that a helm operator for OLM is not the way we want to go. As such, if we can find people to proper support an OLM operator for external-secrets, we will be more than happy to incorporate it again - as long as this is not a forcing function of the current release process.

As such, I am closing your contribution. Sorry about all of the fuss this might have caused 🙏 🙇

[raelga]

It's OK @gusfcarvalho, let us know if you need any help with the non-helm based operator. We are more than happy to help.

In the meantime, we are probably going to internally deprecate the usage of OLM operator and deploy directly with Helm.