Add AI-DLC Vision Document as a Discovery output

[shreelak]

What & why

Makes the AI-PLC → AI-DLC handoff native. Today AI-PLC's Discovery Document is a Working-Backwards artifact; AI-DLC's Inception phase expects a Vision Document (per docs/writing-inputs/vision-document-guide.md in awslabs/aidlc-workflows). This PR adds a step that emits that Vision Document directly, so a PM can finish Discovery and hand a build team an input AI-DLC consumes as-is.

Changes

• New rule discovery/vision-document-generation.md — reshapes existing Discovery content (PR/FAQ, pain points, product strategy, GTM, prototype specs) into AI-DLC's Vision Document schema. Auto-infers Risks and Open Questions from context (no new PM questions) and flags them for review. Deliberately excludes tech/stack details — those belong in AI-DLC's Technical Environment Document, authored in the build workspace.
• core-workflow.md —
  • Runs Vision Document Generation automatically after Go-to-Market on every path (final Discovery output).
  • Adds Entry Point 4: generate a Vision Document from an existing discovery-document.md without re-running Discovery.

Testing

Ran the full Discovery workflow end-to-end on a sample use case (a food-discovery app). Verified the generated vision-document.md:

• conforms structurally to AI-DLC's vision-document-guide.md (all required sections/subsections, no unfilled placeholders);
• correctly excludes technology/stack choices from the source Discovery Document (abstracts vendors to capability-level integration points);
• produces populated MVP IN/OUT scope tables and useful pre-declared Open Questions for AI-DLC Requirements Analysis.