feat: Add initial support for a kafka-based ingest path

[benclive]

What this PR does / why we need it:
Adds write path using Kafka. There are two components: the partition ring which is updated by the ingesters and read by the kafka tee, and the kafka tee itself which is instantiated and managed in the distributor.

• Kafka supports use of the Partition ring to assign partitions based on the number of active ingesters
  • I've attached the ring to the rf1 ingesters for now - we can easily move it to the regular ingester if preferred
• Two new configs: One for the partition ring params, one for enabling this kafka ingest path. We can add more kafka params here.

I put the code under two packages in the ingester-rf1 package to keep it separate, but definitely happy to reorganise!
Most of the config setup is copied from Mimir, including injecting the config at runtime into the ingester config so it can start up successfully.

• Ring management page shows the active status and metrics report success.
• I had to add an small override to enable this to run locally when the hostname does not match the ingester name regex - not sure what the best option is here!

[cyriltovena]

LGTM