Bootstrap the Agentic Runtime research/ideation process

[rkoster]

Summary

Opens the working group's crowd-sourced research phase and puts the process itself up
for review before content arrives at volume.

This PR adds:

• IDEATION.md — the ideation brief: a four-phase roadmap (research → workshop → emergent
  themes → POC/RFC). Only Phase 1 (research intake) is built; later phases are described,
  not scaffolded.
• README.md, CONTRIBUTING.md, CODE_OF_CONDUCT.md — orientation, contribution
  workflow, and conduct.
• research/ — a flat intake directory with the frontmatter schema (research/README.md),
  a copy-me TEMPLATE.md, and one worked example note (cf-runtime-gaps.md).
• .github/ — CODEOWNERS routing to the WG leads, a PR template, and a lightweight CI
  check that validates each note's frontmatter, filename, and required sections.

Design notes

• Themes are not predefined. Notes are organized flat and tagged; the workshop clusters
  tags into themes so the agenda emerges from what people contribute.
• Lightweight by design. Research notes are low-risk; CI + a quick tech-lead check is
  the merge bar. Substantive review of ideas happens at the workshop.
• Real seed notes will follow as separate PRs that exercise this very process.

Feedback wanted

Please review the process (the brief and contribution flow) — this PR exists to gather
that feedback before the research window opens.

[beyhan]

LGTM (I have only a small comment on the pr template)

[wayneeseguin]

This mirrors a lot of how I tend to think about this kind of workflow, so the structure looks like a good starting point to me.

One question: should we consider separating the lifecycle stages more explicitly? For example, ideas, research, and proposals.

Right now, ideas and research seem somewhat combined, which may be fine. I’m wondering whether a little separation plus cross-linking would make the repo easier to browse as it grows: browse raw ideas, follow them to related research, then see which proposals came out of that work.

This may be overkill, but it was the main thought that came to mind...

[jorbaum]

LGTM. Good idea to have a linter for keeping a similar structure.

[rkoster]

should we consider separating the lifecycle stages more explicitly? For example, ideas, research, and proposals. … browse raw ideas, follow them to related research, then see which proposals came out of that work.

Good call — adopted in 3829430. This PR adds an ideas/ lane as a low-barrier front of the funnel, kept distinct from sourced research notes, with cross-links between the two.

One refinement on the shape: rather than a strict ideas → research → proposals conveyor, I've framed research as a reusable layer — a single research note can back several ideas — and it's the idea that carries forward: one with enough research behind it can grow into a proposal. So the browse path you described still holds (raw ideas → the research backing them → the proposals they became), just with research shared across ideas rather than owned one-to-one.

proposals/ itself I've deferred — that's phase 4 (per-theme POC/RFC work that comes out of the workshop), so I'd rather add the directory when we reach it than seed an empty stage.

Thanks for the nudge; it sharpened the model.