[CT-87] A single /health endpoint doesn't distinguish liveness from readiness

[mftee]

Overview

contract/src/handlers/health.rs and contract/src/health.rs provide one /health endpoint (CT-60), but liveness (is the process running) and readiness (can it currently serve traffic, e.g. is Redis/Horizon reachable) are different questions an orchestrator needs separately — a Redis blip currently either makes the whole service look 'down' for restart purposes, or is invisible, depending on how the single endpoint is wired.

Tasks

• Split into /health/live (process is running) and /health/ready (dependencies reachable), or clearly document why a single endpoint is sufficient
• Add tests for each distinguishing a dependency outage (ready=false, live=true) from a process-level problem

Acceptance Criteria

• Liveness and readiness are distinguishable via the health endpoint(s), each independently tested

Notes for Contributors

Comment below to be assigned.