Enhance Mocked vLLM App with Dynamic Metrics for Autoscaling Efficiency

[kr11]

🚀 Feature Description and Motivation

I propose adding a 'metrics' endpoint in docs/development/app/app.py that returns Prometheus-style metrics results.

By mocking the impact of replica changes on metrics, where app.py will detect the replica number and return inversely proportional metric values, we can improve the testing and development of autoscaling policies.

We further add an unitest and enrich README.

Use Case

No response

Proposed Solution

related PR: #116