Improved metrics semantics. Added metrics context enrich via metadata. Introduced mcp tool

[LorenzoTettamanti]

Changes

• Added semantic module to better represent metrics and improve the core structure.
• Moved old service discovery module in the core library
• Added metadata enrichment to metrics in user space
• Improved consumer structure with metadata enrichment
• Available metadata:
  • tgid
  • command
  • runtime
  • container_name
  • container_id
  • pod_name
  • namespace
• Added SSL_read and SSL_write event counter using uprobes and uretprobes attached to libssl.so
• Updated grafana dashboard in Examples/run-with-docker : Added container_name aggregations
• Added helper function in core library to easily attach uprobes and uretprobes
• Minor bug fix and refactoring

Tests (including existing bugs)

image:
lorenzotettamanti/cortexflow-metrics:otel-metrics-23

Tests performed

1. Ran cortexflow-metrics service alongside a toy application. The complete pipeline features frontend+backend + cortexflow-metrics + otel_collector + prometheus + grafana dashboard. The Grafana dashboard template is included in the PR in the folder Example/run-with-docker . The template already perform aggregations using the container-image
2. MCP tests using opencode with the same setup (frontend+backend + cortexflow-metrics + otel_collector + prometheus + grafana dashboard). MCP returns correct results for all the exposed tools

Notes (Must read)

• Default grafana port is 3000
• Metadata injection is not fully finished yet. but the concept seems promising. Expect some bugs or undefined behaviour during the testing phase. Sometimes the service discovery fails and the container_name metadata returns null . I'll try to fix them in the next update (v0.1.6)