ci: add redis cache testing to nix environment

[GEET3001]

Description

Resolves #1978

Notes to the reviewers

• Creates start-redis-single and stop-redis-single Nix scripts for standalone Redis testing (port 6379).
• Creates start-redis-cluster and stop-redis-cluster Nix scripts for 3-node Redis testing (ports 7001-7003).
• Updates redisShellHook to export the exact environment variables expected by cdk-axum for both topologies.
• Creates a test-redis recipe in justfile that spins up both Nix-managed environments and verifies cdk-axum with the redis feature enabled.
• Creates a new redis-tests CI job in GitHub Actions that runs these tests in the regtest environment, mirroring the existing postgres-tests job.

Suggested CHANGELOG Updates

ADDED

• Local Redis cluster and single-node testing capabilities natively within the Nix flake (no Docker Compose required).
• Dedicated CI job (redis-tests) to validate the Redis cache backend in cdk-axum.

Checklist

• I followed the code style guidelines
• I ran just quick-check before committing

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 71.59%. Comparing base (feaf823) to head (9e22e37).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2024      +/-   ##
==========================================
- Coverage   71.60%   71.59%   -0.01%     
==========================================
  Files         356      356              
  Lines       74106    74106              
==========================================
- Hits        53065    53059       -6     
- Misses      21041    21047       +6     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[GEET3001]

All the CI checks are failing with No space left on device (os error 28). The GitHub Actions runner disk is full.

[cdk-bot]

Verified findings approved for disclosure:

• Register the Redis cleanup trap before starting Redis (medium) - If cluster startup fails, just test-redis leaves the previously-started single-node Redis process running, causing local resource leaks and possible port conflicts or stale state on subsequent test runs.

[thesimplekid]

The new test-redis recipe starts Redis, but the cargo test -p cdk-axum --features redis --lib suite does not currently exercise a live Redis cache. The full test list only includes config parsing/default tests plus unrelated handler/router/ws tests; nothing instantiates HttpCache::from_config with Backend::Redis or calls set/get through the Redis backend. That means this CI job can pass even if single-node or cluster Redis cache operations are broken.