In #8448 we decided to defer pytest worker parallelization until it is needed.

I think a good way to do this would be just to template the BIND9 configuration in DNSServer._configure_bind based on what pytest nodes are configured.

For example, example.com would be example.com.{worker_id}.wtf, matching the existing domain generation pattern. Same with any mock recursive records in the mock-recursion RPZ zone.

Then we could simply change the tests as they are written now, to use context.get_domain(domain) rather than domain.

We might need a templating language in order to do this cleanly.