Optimize and batch merge pipelines list splits queries

[guilload]

Description

• Add node_id column to splits table
• Populate node_id column from splits.split_medata_json column
• Add hash index on splits.node_id column
• Batch list splits query when new spawning new indexing pipelines

How was this PR tested?

• Updated unit tests
• Ran this SQL query on Cicada which did not return null or empty values

[guilload]

@fulmicoton, I still have to add at least three unit tests in indexing_service.rs and merge_pipeline.rs, but this PR is ready for review.

[fulmicoton]

Is there a case where we want to call this with None? What is it?

[guilload]

So I covered the spawn_pipelines branch, which is the one apply_indexing_plan uses, but I did not handlespawn_pipeline, which I think is used by the local CLI ingest. This is the reason for using an option.

I will refactor the code a little bit to eliminate the option. With some tweaking, spawn_pipeline can use spawn_pipelines under the hood.

[fulmicoton]

Suggested change

	UPDATE
	// Split metadata have been stable for quite a while, so we
	// allow ourselves to do this, but please reader of the future,
	// do not blindly reapply this pattern.
	UPDATE