Implement agentic mobility and MRO updates

[ididhalfazan]

Sprint 3: Full Agentic Upstream (MRO + Mobility)

Acceptance Criteria

• 3.1 MRO source upstream + validation tests ✅
• 3.2 Shared LLM provider (Groq/Bedrock/OpenAI/Ollama) ✅
• 3.3 Mobility pipeline tests (NL → UE tracks) ✅
• 3.4 Scenario presets + --preset CLI flag ✅
• 3.5 Pinned requirements-agentic.txt ✅

Changes

New files:

• radp/agentic/llm/provider.py — Unified LLM provider
• tests/test_validation.py — MRO validation tests
• tests/agentic_mobility/ — Pipeline & preset tests
• radp/digital_twin/agentic_mobility/defaults/ — Urban/suburban/rural presets
• requirements-agentic.txt — Pinned dependencies

Modified files:

• MRO LLM provider → re-export shim (backward compatible)
• Mobility llm_client.py → delegates to shared provider
• Parameter agent & end-to-end example wired for presets

Key Points

• Unified provider supports 4 backends, swappable via config
• Full pipeline tested with deterministic outputs
• Scenario presets generate distinguishable results
• Zero breaking changes, all tests passing
• Dependencies locked to prevent breakage

Note: Mobility module at radp/digital_twin/agentic_mobility/ (non-blocking path decision)