Fix small false-positive for ViewBinding

[tasomaniac]

I recently opened #710 about suppressing view binding check. I had to suppress since there was a false positive. I looked into our project more and figured out what was wrong.

This is coming from developer-menu module where we have different layout for debug and release versions. In case you want to have different layout in different build-types, you have to separate them. But then you can use them in main source set when accessing. ModuleCheck does not account for this valid (but uncommon) case.

This PR fixes the issue in most simplest way with a test case. Let me know if this makes sense. I'm not sure if I missed anything. If you have concerns about the change, I hope the test case will be helpful.

[tasomaniac]

There is insufficient memory for the Java Runtime Environment to continue.
Native memory allocation (mmap) failed to map 117440512 bytes for Failed to commit area from 0x0000000090000000 to 0x0000000097000000 of length 117440512.

[RBusarow]

There is insufficient memory for the Java Runtime Environment to continue.
Native memory allocation (mmap) failed to map 117440512 bytes for Failed to commit area from 0x0000000090000000 to 0x0000000097000000 of length 117440512.

Yeah...

I moved the test matrix out of GitHub Actions and into dynamic test factories. It's great locally, but the memory issues with TestKit get multiplied and that's too much for the free Windows GHA runner to handle. I'm hoping to move those tests over to an EC2 instance soon. In the meantime, the tests usually pass on the first retry. :)

[RBusarow]

Thanks for this! I understand the problem.

Ultimately, this problem affects all Android resources. I'll add something later to handle those cases, but this fix is perfectly fine and it unblocks you, so LGTM. :)