note that we also have uuid generation on scope.propagation_context now and to make that lazy we first need to encapsulate propagation_context in a class and not just a raw dict as currently implemented.

Without also making those lazy, this will not really have a performance impact but we can merge it anyway.

Is there a reason to not land this pull request right now?

we can merge but without the prop context changes it won't have full impact, and that might conflict with the v2 refactor, I will let @sentrivana and @antonpirker decide

I will refactor the propagation context first and then merge this change

This has found its way into the codebase with the refactoring of the propagation context: #2970

UUIDs in tracing.py are still always generated, as #2970 only made the generation lazy for propagation contexts. Is this intentional?