Experimental route evidence and MeshMapper scope context

[n30nex]

The experimental server is based on the released v2.0.0 main/dev commit 0015430. Its remaining delta supplies cached MeshMapper regional catalogue context, node-list default scope, and exact-route evidence/cursors. My Atlas and Topology remain frontend features; Collector intake runs in a separate private service and database, with no collector package, table or migration in this branch.

The Canadaverse preview runs this exact server head 8c7fbb8 (2.1.0-n30nex.2) with web 7e139d5 (2.1.1-n30nex.1). It uses the 001 database baseline with the legacy database and verified recovery copies intact. At the user's request, effective retention matches released defaults: raw 7 days, summaries 90 days, observer telemetry 31 days and routes 14 days.

Validation: native ARM64 build/vet and 753 server tests with real PostgreSQL; two opt-in backup/export tests skipped. Published-head CI/CodeQL passes. The paired web passed 1,259 tests; 31 post-deployment API checks, exact public assets/source verification and desktop/French phone journeys passed. The earlier 3,200-input replay is historical evidence, not a fresh benchmark for this head.

Current audit · Integration guide.

Draft for later acceptance. Production was not modified; no stable release or review request. Daily compatibility automation remains paused.

[MrAlders0n]

Closing this one out. The saved-route prefix issue from #183 is fixed on dev in #198: UpsertKnownRoute now refreshes hash_prefix on conflict, so evidence follows the route's current hash width. This branch also conflicts with the 2.0 baseline, so it can't land as-is. Thanks for digging into this.