epic: Refactor research domain around Track aggregate root

[Linjie-top]

Summary

Refactor the research domain to treat ResearchTrack as the aggregate root instead of loosely-attached track_id fields across
many tables and services. This epic tracks semantic refactors and read-model consolidation without doing a destructive DB rebuild.

Goals

• Make ResearchTrack the explicit aggregate root in code and docs.
• Consolidate how track-scoped data (memory, feedback, feed, runtime) is read and injected.
• Introduce track-centric read models/services so callers do not have to manually stitch track_id-scoped tables.

Non-goals

• No wholesale schema rewrite or dropping existing tables.
• No major UI redesign; only internal refactors and small UX polish where it directly follows from the new model.