[Jobs Match Making] Design Redis data model for matching metadata

[aldbr]

Description

Design how job scheduling requirements and node characteristics map to Redis data structures. This is the most critical architectural decision for performance.

Consider the query patterns needed for matching (which data needs to be read per match?), memory usage at 10M jobs, atomicity requirements.

Explore multiple options and document trade-offs in a design document.

Definition of Done

• Design document (markdown) committed to the repo
• At least 2 alternative data models described with pros/cons
• Memory estimate at target scale (10M jobs, 1000 sites)