feat: rust-observability skill -- tracing, OpenTelemetry, Prometheus for Rust services

[AlexMikhalev]

Parent Epic: #62

What

New skill: rust-observability covering production monitoring and observability for Rust services.

Skill Content

1. Structured Tracing -- tracing crate setup, span context, #[instrument] attribute, field conventions
2. OpenTelemetry Integration -- tracing-opentelemetry subscriber, span export to Jaeger/OTLP, context propagation across HTTP boundaries
3. Prometheus Metrics -- metrics crate with metrics-exporter-prometheus, histogram/counter/gauge patterns, endpoint setup
4. Per-Request Context -- Request ID propagation, correlation IDs in traces and logs, HTTP header extraction
5. Error Reporting -- Structured error spans, error rate metrics, alerting thresholds
6. Log Filtering -- EnvFilter configuration, per-crate log levels, production vs development configs

Source Reference

Section 13 (Production Monitoring and Observability) from RUST_SYSTEM_PROGRAMMING_BEST_PRACTICES.md

Disciplined Engineering Alignment

Phase	Skill	Observability Relevance
Research	disciplined-research	Identify observability gaps in current service; map existing log/metric/trace coverage; document SLI/SLO requirements
Design	disciplined-design	Specify tracing span hierarchy; define metric names/labels/buckets; design log level strategy per environment
Implementation	disciplined-implementation	Add tracing subscriber first, then instrument hot paths, then metrics, then OTel export -- each as separate step
Verification	disciplined-verification	Verify spans appear in collector; confirm metric cardinality within bounds; test log filtering at each level
Validation	disciplined-validation	Validate observability under production load; confirm alerts fire correctly; stakeholder sign-off on dashboard

Key integration point: The devops skill should reference rust-observability for Rust service deployment monitoring requirements.

The SKILL.md should include:

• A "Disciplined Observability Checklist" mapping each V-model phase to specific observability tasks
• Integration examples for axum middleware (tracing + metrics)

Why a New Skill (Not Extension)

Observability is a distinct cross-cutting concern for any production Rust service. It's not specific to development idioms or performance optimization.

Relevance to Terraphim

terraphim-server runs as a persistent service. Structured tracing and metrics are essential for production operation and debugging.

Acceptance Criteria

• skills/rust-observability/SKILL.md created with full content
• Includes axum middleware examples for tracing and metrics
• Includes Cargo.toml dependency snippets
• Covers both development (pretty logs) and production (JSON/OTLP) configurations
• Disciplined observability checklist included
• Cross-references devops skill