Add SyncV2 skeleton and add migration to store serialized components

[ChrisPenner]

Overview

• TODO: Before deploying ensure you've created BOTH the migration queue table and the serialised component tables.

Adds the API and handlers for syncv2 pulls, as well as code to run a migration to backfill storing serialized rows for all existing code stored in Share (this'll take a while to run I think 😓 )

Implementation notes

• Adds new syncv2 'download-entities' endpoint which takes a HashJWT for a given root and will send a stream of all dependencies of that root.

Test coverage

I've done some manual testing, but this is going to remain experimental until we've bullet-proofed it in UCM anyways and shouldn't affect anyone who's not using an experimental UCM build

Loose ends

I have branches in progress for these:

• Sort entities in dependency order
• Allow UCM to pass hashes whose dependencies should be omitted from the stream