v0.3.0

This release makes the performance evidence and observability paths operational end to end.

Highlights

OTEL metrics are exported to Prometheus and available through a provisioned Grafana datasource.
Main load runs enforce the committed measured baseline, with positive and negative CI fixtures.
Spike and breakpoint thresholds are centralized; stress and soak workloads enforce endpoint SLOs.
Human-readable summaries include business metrics and the report site charts latency trends.
Observability health checks work on Windows, Linux, and macOS.
Concurrent-reader behavior, soak analysis, and authenticated onboarding are now explicit.