[Feature] Advanced Observability: Deep OpenTelemetry Integration

[tercel]

Problem Description

Current observability is 'span-based' but custom. It lacks native, first-class integration with the full OpenTelemetry (OTel) ecosystem (Logs, Metrics, Traces) across all languages.

Why solve it

Enterprise users expect OTel compatibility for production monitoring. Standardizing this in the protocol ensures consistent 'AI-aware' tracing.

• Production Readiness: Providing standard tools for monitoring and debugging.
• Interoperability: Integrating with existing observability stacks (e.g., Prometheus, Grafana, Honeycomb).
• Consistency: Ensuring OTel traces look the same across all SDK implementations.

How to solve it

1. Trace Context Mapping: Map apcore context and traces to W3C Trace Context and OTel semantic conventions.
2. OTel Mapping Spec: Specify how UsageCollector and ErrorHistory map to OTel Metrics and Logs.
3. Reference Middleware: Provide reference OTel middleware in Python, TypeScript, and Rust SDKs.
4. Documentation: Update docs/features/observability.md with OTel-specific integration guides.