Add Phase 1 chat orchestration foundation

[paullizer]

Summary

Introduces the generic Phase 1 foundation for coordinating SimpleChat capabilities per submitted chat turn. Grounded image generation is the first proving profile, while the planner and metadata contracts remain output-neutral.

This PR intentionally targets feature/orchestration so the phased capability can be integrated and tested without exposing an incomplete orchestration workflow in Development.

Changes

• Adds deterministic direct and coordinated turn plans without an additional planning-model call for simple requests.
• Captures a compact submitted-control snapshot for selected agents, document actions, documents and scopes, retrieval modes, conversation-linked documents, assigned knowledge, model selection, reasoning effort, and prompt identity.
• Separates inferred evidence requirements from deduplicated source attempts and records source origin, required state, execution dependencies, finalizer dependencies, and policy defaults.
• Integrates orchestration metadata with standard streaming and Analyze/Compare document-action paths, including success, cancellation, and partial-error metadata.
• Adds coordinated finalizer guidance for required source attempts, partial results, conflicts, and unsupported claims.
• Aligns image proposal guidance with the orchestration image-intent detector while keeping chart and analytical requests outside image generation.
• Uses grounded image generation as the initial task profile, including evidence discovery when grounding is requested without a named source.
• Adds feature documentation, release notes, and version 0.250.058.

Validation

• functional_tests/test_chat_turn_orchestration_plan.py: 15/15 passed
• functional_tests/test_image_proposal_pipeline.py: 3/3 passed
• functional_tests/test_chat_streaming_sse_metadata_fix.py: 3/3 passed
• Route policy contracts: 12/12 passed
• Python py_compile: passed
• VS Code diagnostics: clean
• Scoped broken-access-control scan: passed
• git diff --check: clean

Known Follow-Up Work

• Add the shared result/evidence ledger.
• Adapt history, prior citations and cited documents, workspace, web/research, agents, and actions to generic executor contracts.
• Move selected agents beneath a true central finalizer for coordinated turns.
• Add generic scheduling, per-step reconciliation, governed capability discovery, retries, bounded replanning, budgets, and durable resume.
• Add a dedicated selected-image-reference request field; selected workspace images currently arrive as documents and explicit reference intent is inferred from the message.

Refs #1021