IssueTracker

Recently, we've begun seeing lint errors crop up around 3rd party lint rules (in our case from Jake Wharton's Timber). These lint errors imply that lint can no longer find a lint rule in our configuration file from the library.

Here's an example:

```
16:10:19 > Task :repository:mergeInternalDebugUnitTestResources
16:10:19
16:10:19 FAILURE: Build failed with an exception.
16:10:19
16:10:19 * What went wrong:
16:10:19 Execution failed for task ':app:lintInternalDebug'.
16:10:19 > Lint found errors in the project; aborting build.
16:10:19  
16:10:19   Fix the issues identified by lint, or add the following to your build script to proceed with errors:
16:10:19   ...
16:10:19   android {
16:10:19       lintOptions {
16:10:19           abortOnError false
16:10:19       }
16:10:19   }
16:10:19   ...
16:10:19  
16:10:19   Errors found:
16:10:19  
16:10:19   /Volumes/Code/Jenkins/workspace/Faithlife-Android/app: Error: Unknown issue id "BinaryOperationInTimber", found in /Volumes/Code/Jenkins/workspace/Faithlife-Android/app/lint.xml [LintError]
```

Our lint configuration related to that rule since we adopted the library (about two years ago). Initially, I thought it might be related to Gradle because of https://github.com/JakeWharton/timber/issues/316 and https://issuetracker.google.com/issues/145235363. However, after disabling parallel builds (which shoudln't have any affect on our project anyway) and incremental kapt for good measure, the problem persists. I also deleted the gradle dependency cache and ran clean before building, again for good measure.

The difficult part of the problem is that it seems to show up on different machines randomly. Between the two full-time android developers on our team and our build server, we've each seen the problem crop up at different times.

The only thing that seems to squelch the issue is reverting AGP to 3.5.2 (without any other changes!). As far as we can tell, this addresses the issue on every machine we use.

I'm unsure if this interaction is specific to Timber, lint, or some weird interaction of the two. But issues like these are pretty obvious deterrents from making good use of static analysis (our team has even written custom rules!). We can't really work around weird problems like this since the problem seems to exist outside of any changes we've made.

I haven't made a sample reproduction of the problem because, as I've said previously, the problem itself is relatively indeterminate.