Is your feature request related to a problem? Please describe.
In our company, we have a central project in which GKE is managed, but then every application has its own project which gets it's own namespace in that central GKE setup. Now we are trying to integrate with the central logging architecture provided by our central TPS team.

Everything works, but there is a slight problem for us in the autodetection of the MonitoredResource. It detects the central project instead of the project that owns the workload. Because this central team wants to forward logging bills to the projects that are actually doing the request, they enforce us to provide the project id in the resource.project_id field.

I've tried a couple things, without success:

• Use the logging.DetectProjectID setup
• Only setting the project_id field and hope the auto-detection will set the rest (this was a desperate attempt but I wanted to make sure it didn't work before filing this issue)

Describe the solution you'd like
The Java library used by other teams in our company seems to first check for the "GOOGLE_CLOUD_PROJECT" environment variable before going to the metadata server. If we also do this here (at least for the GKE detection) this would solve our issue.

Describe alternatives you've considered
I want a way to override the detected resource.project_id field for log entries.
We could implement a new LoggerOption like logging.OverrideResourceProjectId(projectID string). which would then set the "project_id" label on l.commonResource to the given projectID.

The only other alternative I see for us is to write our own auto-detection setup, which I really dislike, as this feels like something that will change often.

Additional context
I would be happy to implement this feature if maintainers agree that it would fit.