feat(rmi): add curated Roads Management Insights (RMI) agentic skills suite

[n0531m]

Summary

This PR introduces the curated Roads Management Insights (RMI) Agentic Skills Suite (v1.0.0) under roads_management_insights/.agents/skills/.

This suite equips AI developer agents (such as Jetski, Antigravity, Claude Code, etc.) with grounded domain intelligence, API wrappers, BigQuery SQL libraries, visualization tools, and route setting strategies to architect and operate RMI solutions.

---

Curated Skills Included (13 Skills)

1. Foundations & Routing

• api-roads-v1: Roads API v1 client for snapping coordinates to Google road segments and speed limits.
• api-routes: Routes API v2 client with TRAFFIC_UNAWARE baseline calculations and polyline extraction.
• api-roadsselection: Roads Selection API client for SelectedRoute creation, listing, updating, and deletion.
• api-analyticshub: Analytics Hub API client for discovering and subscribing to shared RMI datasets.

2. RMI Specifications & Contracts

• rmi-data-foundation-bigquery: Schema definitions, DDLs, and partition strategies for foundation BigQuery tables (routes_status, historical_travel_time).
• rmi-data-realtime-bigquery: Schema contracts and stream translation patterns for real-time telemetry tables (recent_roads_data).
• rmi-data-realtime-pubsub: Pub/Sub streaming message contracts and ingestion architectures.

3. Analytics, Route Setting & Operations

• rmi-routesetting: Canonical route setting strategies:
  • SIMPLE_ORIGIN_DESTINATION: Dynamic baseline setting via TRAFFIC_UNAWARE.
  • SINGLE_ROUTE_UNIFORM_INTERMEDIATES: Pure bash + jq adaptive existing-vertex selection with distance-aware spacing ($\ge 200\text{m}$) and zero synthetic points.
  • MATCH_AND_SPLIT_BY_ROAD: Pure bash + jq road-segment midpoint pinning via snapToRoads clusters and adaptive 25-waypoint limit capping.
  • BUS_ROUTE_MONITORING: Stop-to-stop transit hop sequencing with BigQuery window function analytics.
  • BYO_POLYLINE: Open GIS vector ingestion with pre-flight geometric overlay feasibility assessment and intent-based recreation guidance.
• rmi-sampledatasets: Sample dataset catalog, schema extraction patterns, and string-only attribute casting guidelines.
• rmi-sql: Parameterized BigQuery SQL query library with Job ID headers, partition bounding, and window analytics.
• bigquery-geospatial: Geospatial analytics, geometry functions (ST_INTERSECTS, ST_CENTROID), and GeoJSON converters.
• bigquery-practices: Best practices for cost estimation, partition pruning, and schema optimization.
• rmi-dataviz: Visual layout guidelines and dashboarding patterns.

---

Manifest & Documentation

• Includes .bundle-manifest.json declaring skill versions and checksums.
• Includes README.md and high-contrast architectural dependency diagram under diagrams/rmi-core_dependency_graph.png.

---

Verification

• All skills validated against automated unit test suites (tests/test_unit.sh) and agent evaluations (evals.json).
• Monorepo clean build verified with 100% test pass rate.