Support distributed trace propagation #122

Description

Allow applications behind instrumented infrastructure to join distributed traces and propagate context to upstream requests.

Main uses the tracing facade for request log messages, but the public core/proxy APIs have no distributed trace context extraction/injection or span lifecycle integration.

Steps to reproduce

1. Inspect the request logger and single-request proxy boundary.
2. Look for inbound trace-context parsing and outbound propagation in the public module surface.

Expected behavior

Incoming trace context is validated, associated with request processing and propagated according to a documented integration policy.

Impact

Request logs alone cannot correlate a request across instrumented services.

Proposed fix

Define portable trace context integration and adapter hooks, then document the supported propagation format, span boundaries and optional exporter setup.

Done when

• Supported inbound trace context is parsed/validated and attached to request work.
• Outbound requests propagate the chosen context under an explicit policy.
• Request/error/async lifetime boundaries are documented and covered.
• Exporter/backend integration remains optional and portable core gains no runtime dependency.

Related

• Parent: Complete portable request observability #118.
• Structured logs: Provide consistent structured logs across adapters #119.
• Generic timing is separate: Add a shared request-timing collector and opt-in attachment middleware #446.
• Built-in request logger
• Current single-request outbound interface
• Core public module surface