Karafka & Waterdrop: support different configurations per topic

[Drowze]

NOTE: this pull request is blocked until #4876 and #5120 are merged

---

What does this PR do? Supports different karafka/waterdrop configurations per topic.

Motivation: Most of our Karafka consumers simply iterate through a kafka batch of messages and process them individually. Some of our consumers however have a some logic to pre-process a batch (e.g.: split in smaller batches based on some logic) to process more efficiently later. In such cases we want to selectively disable Datadog's Waterdrop/Karafka tracing and handle tracing manually ourselves.

Change log entry

Additional Notes:

How to test the change?

Automated tests provided.