Make uptime-ready-check.sh honor READY_FAIL_ON_DEGRADED

## Summary

`/api/v1/ready` returns HTTP 200 when status is `degraded` unless `READY_FAIL_ON_DEGRADED=true`. `scripts/uptime-ready-check.sh` treats `degraded` as OK, so monitors will not page on optional-dep failure.

## Context

- `src/voxforge/infrastructure/observability/health.py`
- `scripts/uptime-ready-check.sh`
- `docs/deployment/uptime.md`

## Acceptance criteria

- [ ] Optional flag (e.g. `FAIL_ON_DEGRADED=true`) fails on `status=degraded`
- [ ] Default behavior unchanged
- [ ] Docs show both monitor recipes
- [ ] Small test or documented fixture for JSON parsing

Start with scripts/uptime-ready-check.sh and compare its handling of the /api/v1/ready response with READY_FAIL_ON_DEGRADED in src/voxforge/infrastructure/observability/health.py. Add the optional degraded-status failure behavior while preserving the default, then update docs/deployment/uptime.md with both monitor recipes and add a small JSON-parsing test or documented fixture.

Written by the indexing model from the issue text.