feat(skills): add VidXP agent workflows

[tulayha]

Summary

• add a public vidxp-ingest-video skill for local-path and browser-handoff ingestion with automatic indexing
• add a public vidxp-find-video-evidence skill for grounded search, keyframe inspection, and clip delivery
• declare ChatGPT-facing metadata, implicit invocation, and the VidXP MCP dependency for both workflows
• link the downloadable skill sources from the product README

Why

Natural requests such as “index this video” and “find where Harry appears” should select VidXP and follow its native MCP workflow without requiring users to name individual tools or reconstruct timestamp-to-clip sequences.

User impact

The skills teach ChatGPT and Codex to discover the VidXP connector, choose transport-appropriate ingestion, poll the authoritative durable operation, and return inspectable evidence. They are stored as public source packages under skills/; plugin packaging remains separate.

Validation

• both skill folders pass the official quick_validate.py validator
• ChatGPT metadata and MCP dependency declarations parse successfully
• README links resolve
• git diff --check passes

BEGIN_COMMIT_OVERRIDE
feat(agents): add guided video ingestion and evidence workflows
END_COMMIT_OVERRIDE