test(python): qualify measured local first use

[jeremydixon22]

Summary

• report first-citation, durable-receipt, restart-recovery, and completion timings from the real local quickstart
• run wheel and sdist installs through citation, receipt-before-dispatch, restart, second-process replay, inspection, and safe reset
• require that measured evidence in every Python 3.10-3.12 Linux, macOS, and Windows matrix cell
• allow the platform matrix to run before a published upgrade baseline exists while keeping the upgrade gate conditional and explicit

Verification

• full Python runtime gate: 195 tests, 79.1% combined line/branch coverage, strict mypy across 76 source files, wheel/sdist/server-extra checks
• clean wheel and sdist quickstarts completed locally with first citations in under one second
• platform-matrix verifier positive and over-budget rejection cases
• public-export reproducibility, JSON, link, secret, ownership, and history scans

Public information

Yes. This adds user-facing quickstart timing fields and versioned clean-install evidence fields. It does not publish a package, promote runtime maturity, or expose private planning material.