fix(ingest): Do not require platform_instance for stateful ingestion

[asikowitz]

Following the changes from #6795, I believe that get_platform_instance_id can now return None. See its only usage:

In StatefulIngestionSourceBase._get_last_checkpoint

                try:
                    platform_instance_id = self.get_platform_instance_id()
                except NotImplementedError:
                    pass
                else:
                    last_checkpoint_aspect = self.ingestion_checkpointing_state_provider.get_latest_checkpoint(  # type: ignore
                        pipeline_name=self.ctx.pipeline_name,
                        job_name=job_id,
                        platform_instance_id=platform_instance_id,
                    )

Note that in that PR, get_latest_checkpoint was configured to accept an optional platform_instance_id (and also note that in the not implemented case, None is implicitly passed?)

Since we still want to keep implementations of get_platform_instance_id for backwards compatibility, I allow it to return None for when you sometimes have a platform_instance, and other times don't, like with Kafka.
Also removes any validation around null platform_instance that I could find.

EDIT: Tested the following recipe

source:
  type: kafka
  config:
    connection: 
      bootstrap: localhost:9092
      schema_registry_url: http://localhost:8081/

Checklist

• The PR conforms to DataHub's Contributing Guideline (particularly Commit Message Format)
• Links to related issues (if applicable)
• Tests for the changes have been added/updated (if applicable)
• Docs related to the changes have been added/updated (if applicable). If a new feature has been added a Usage Guide has been added for the same.
• For any breaking change/potential downtime/deprecation/big changes an entry has been made in Updating DataHub

[hsheth2]

These changes aren't strictly necessary, since mypy allows an override method to narrow the return type