feat: implement metric backend

[StarpTech]

Motivation and Context

This PR instrument the Router with OpenTelemetry Metrics. All metrics are exposed through the PrometheusExporter on port http://127.0.0.1:8088/metrics (configurable) and sent to the otelcollector metrics endpoint.

The PR introduces built-in support for Prometheus. Data is exposed at http://127.0.0.1:8088/metrics. We export the default Go and Process metrics. In addition, we export (R.E.D) metrics related to incoming GraphQL traffic:

• router_http_requests_total: Total count of incoming requests
• router_http_response_content_length_total: Total bytes of incoming requests
• router_http_request_content_length_total: Total bytes of outgoing responses
• router_http_duration_milliseconds: End-to-end duration of incoming requests in milliseconds (histogram)
• router_http_in_flight_requests: Number of in-flight requests

All metrics are tracked along the following dimensions:

• operation_name
• operation_type
• status_code
• federated_graph_name
• config_version

This enables you to answer the following questions:

• What is the error/success rate of my router or a specific operation?
• How is the performance of my router or a specific operation?
• What is the average request/response size of a specific operation?
• How much traffic went through a router instance?

During this work, I found dealing with net/http handlers and custom gin middleware very cumbersome to manage. I used the time to migrate to chin which is 100% compatible with the std library. This is also an important step towards custom middleware support in the router. I refactored the GraphQL handlers for easier testing as well.

TODO

• Test Helm Charts

• Tests or benchmark included
• Documentation is changed or added on https://app.gitbook.com/
• PR title must follow conventional-commit-standard