Extend gomock to allow passing an source_importpath instead of library when operating in source mode

[ramenjosh]

What type of PR is this?

Feature

What does this PR do? Why is it needed?

This PR adds a new parameter, importpath to the gomock rule. This parameter is optional, and allows a gomock target running in source mode to avoid depending on a go_library. As far as I can see, the only use for the go_library in source mode is to extract an importpath.

I've also extended the test cases to test the different modes gomock can operate in, and made sure there's a link to gomock on the top level README.

Which issues(s) does this PR fix?

Fixes #3349

[fmeum]

This looks very reasonable to me - it's a very simple way to get by without a potentially annoying dependency edge.

[linzhp]

I am afraid the attribute name importpath would confuse Gazelle: now there are two targets with the same importpath value. How about calling it source_package?

[ramenjosh]

I think calling it source_package would cause confusion with package, maybe source_importpath would be a good middle ground?

[ramenjosh]

There's actually also self_package, which has this description:

self_package: the full package import path for the generated code. The purpose of this flag is to prevent import cycles in the generated code by trying to include its own package. See [mockgen's -self_package](https://github.com/golang/mock#flags) for more information.

Perhaps we can avoid introducing a new parameter and just use this one?

[ramenjosh]

Hmm, on second thought it's probably better to avoid overloading the parameter given it has special meaning to mockgen itself. Have gone with source_importpath for now.

[ramenjosh]

@fmeum @linzhp can you please take another review? There's still one part of the CI that seems to be failing and I'm not super sure why, any tips on what might be causing it?

[fmeum]

@ramenjosh To be honest, I have no idea what is causing that failure. Could you amend and force-push your branch to trigger a new CI run? It may just be a flake.

[therve]

@fmeum Can we revive that change? The issue is still present. Thanks.

[fmeum]

Sure, I can review this if someone can resolve the conflicts. The flaky tests should be gone now.

[therve]

Thanks, created #3822.