feat: implement real-time prometheus metrics for experiment pods

[mihir-dixit2k27]

This PR implements real-time Prometheus metrics for chaos experiment pods, addressing the existing TODO in pkg/telemetry/otel.go.

Currently, observability is limited to post-experiment ChaosResult CRDs. This change allows SREs to monitor chaos injection status and impact (e.g., litmuschaos_experiment_injection_count) in real-time via Grafana while the experiment is running.

Key Changes:

• Telemetry: Added a Prometheus exporter in pkg/telemetry/otel.go that initializes an HTTP server on port 8080.
• Instrumentation: Instrumented the pod-delete experiment (chaoslib/litmus/pod-delete/lib/pod-delete.go) as the pilot implementation.
• Dependencies: Updated go.mod and go.sum to include the OpenTelemetry Prometheus exporter.

I have included a 5-second graceful shutdown delay at the end of the experiment logic. This ensures that Prometheus has a sufficient window to scrape the final "Verdict" and "Status" metrics before the pod is evicted and the process terminates.

[mihir-dixit2k27]

Thanks for the review!

1. Go Version & Compatibility: I have updated go.mod to use Go 1.22.0 and ran go mod tidy to ensure compatibility with the rest of the stack (Chaos Operator/Runner).

2. Verification: Metrics Output Since a full Prometheus UI isn't available in this local environment, I validated the exporter by initializing the InitMetrics function and querying the localhost endpoint.

Raw Output Verification: The screenshots below confirm the endpoint is active and serving OpenTelemetry-compliant metrics.

Runtime & Process Metrics:

OpenTelemetry Metadata (Target Info):

Key Metrics Observed:

Runtime Stats: go_goroutines and go_memstats_* are being collected, confirming the runtime hook.

OTel Identification: The target_info metric (visible in the last screenshot) confirms the telemetry_sdk_name="opentelemetry", proving the OTel SDK is successfully registered.

Ready for the next round of review!