Add AI-assisted contribution guidance and strengthen verification expectations

[niels9001]

Fixes

N/A - documentation/process improvement, no tracking issue.

PR Type

• Bugfix
• Feature
• Code style update (formatting, renaming)
• Refactoring (no functional changes, no API changes)
• Build related changes
• Documentation content changes
• Other (please describe):

Description

WinUI's contributor docs had no guidance for AI/agent-assisted contributions, which are increasingly common and, when unmanaged, flood maintainers with low-context or duplicate PRs. Other major Microsoft repos (notably microsoft/TypeScript and microsoft/typescript-go) have addressed this with explicit policies; WinUI had a gap. This PR closes that gap, reinforces that authors must verify their own changes, and adds a maintainer command to request that verification.

Current Behavior

• No policy on AI-assisted or autonomous-agent PRs, and no disclosure mechanism.
• No root AGENTS.md to orient coding agents to the repo's contribution rules.
• Two divergent contribution-handling docs: CONTRIBUTION_HANDLING.md at the root (orphaned, nothing links to it) and the canonical docs/external/contribution_handling.md that CONTRIBUTING.md and README.md link to.
• No quick way for maintainers to ask a contributor to manually verify their change.

New Behavior

• CONTRIBUTING.md: new "Use of AI assistance" section (AI tools are welcome, disclosure is required, bulk/agent-driven PRs are not accepted, plus an "Instructions for autonomous coding agents" subsection). New "Verify your own changes" subsection requiring repro/verification steps and a screenshot or short screen recording for user-visible changes. Heading renamed to "Contributing to WinUI".
• .github/PULL_REQUEST_TEMPLATE.md: adds AI disclosure checkboxes, a verification / repro steps field, a "ran and verified the new behavior myself" checkbox, and a "Screenshots / Screen recording" section.
• New root AGENTS.md: agent-facing orientation covering PR acceptance rules, the proposal gate for public API/UI, manual verification expectations, build/test skills, and PR requirements.
• New .github/workflows/needs-verification-command.yml: a maintainer can comment /needs-verification on an issue or PR to post a templated request asking the author to manually verify their change (linking to the "Verify your own changes" section) and apply the existing needs-author-feedback label, which plugs into the repo's stale-bot automation. The message is intentionally neutral and does not call out AI.
• Removes the orphaned root CONTRIBUTION_HANDLING.md; the canonical copy at docs/external/contribution_handling.md remains.

The AI policy language is adapted from the TypeScript repos but tailored to WinUI's process (2 approvals, CLA, feature-proposal gate, squash merge).

Customer Impact

Contributor-facing process/docs plus one maintainer-only automation. No product or runtime code changes. Sets clearer expectations for community contributors and reduces maintainer triage burden.

Regression Potential

• Low risk — isolated change, limited scope
• Medium risk — touches shared components or public APIs
• High risk — architectural or breaking API change

Docs/templates plus a new comment-triggered workflow that is gated to maintainers (OWNER/MEMBER/COLLABORATOR) and reuses an existing label. Verified internal links resolve, no remaining references point to the removed file, and the workflow YAML parses.

How Has This Been Tested?

Verification / repro steps:

1. Confirmed grep finds no remaining references to CONTRIBUTION_HANDLING.md; CONTRIBUTING.md and README.md still link to docs/external/contribution_handling.md.
2. Verified the "Use of AI assistance" and "Verify your own changes" anchors match the header/command links.
3. Validated needs-verification-command.yml parses as YAML and gates on the command + maintainer association; it reuses the needs-author-feedback label already used by .github/policies/resourceManagement.yml.

• I have performed a self-review of my own code
• I have run and verified the new behavior myself (not just relied on tools or AI output)
• I have added tests to cover my changes
• Existing tests pass locally

AI Assistance

• AI coding tools (e.g. Copilot, Claude, Codex, Cursor) were used to author part or all of this change
• I have read and understand every change in this PR, have verified it works, and can discuss and revise it in review

Screenshots / Screen recording

N/A - text and workflow change.