Deprecate Collections in favour of Plugins

Replace Collections with Plugins as first-class citizens in the repo.
With the Copilot CLI v0.409 release making plugins an on-by-default
marketplace, collections are redundant overhead.

## What changed

### Plugin Infrastructure
- Created eng/validate-plugins.mjs (replaces validate-collections.mjs)
- Created eng/create-plugin.mjs (replaces create-collection.mjs)
- Enhanced all 42 plugin.json files with tags, featured, display, and
  items metadata from their corresponding collection.yml files

### Build & Website
- Updated eng/update-readme.mjs to generate plugin docs
- Updated eng/generate-website-data.mjs to emit plugins.json with full
  items array for modal rendering
- Renamed website collections page to plugins (/plugins/)
- Fixed plugin modal to use <div> instead of <pre> for proper styling
- Updated README.md featured section from Collections to Plugins

### Documentation & CI
- Updated CONTRIBUTING.md, AGENTS.md, copilot-instructions.md, PR template
- Updated CI workflows to validate plugins instead of collections
- Replaced docs/README.collections.md with docs/README.plugins.md

### Cleanup
- Removed eng/validate-collections.mjs, eng/create-collection.mjs,
  eng/collection-to-plugin.mjs
- Removed entire collections/ directory (41 .collection.yml + .md files)
- Removed parseCollectionYaml from yaml-parser.mjs
- Removed COLLECTIONS_DIR from constants.mjs

Closes #711

plugins/gem-team/.github/plugin/plugin.json

@@ -6,5 +6,61 @@
     "name": "Awesome Copilot Community"
   },
   "repository": "https://github.com/github/awesome-copilot",
-  "license": "MIT"
+  "license": "MIT",
+  "tags": [
+    "multi-agent",
+    "orchestration",
+    "dag-planning",
+    "parallel-execution",
+    "tdd",
+    "verification",
+    "automation",
+    "security"
+  ],
+  "display": {
+    "ordering": "manual",
+    "show_badge": true
+  },
+  "items": [
+    {
+      "path": "agents/gem-orchestrator.agent.md",
+      "kind": "agent",
+      "usage": "recommended\n\nThe Orchestrator is the coordination hub that coordinates multi-agent workflows, delegates tasks via runSubagent, and synthesizes results. It does not execute tasks directly but manages the overall workflow.\n\nThis agent is ideal for:\n- Coordinating complex multi-agent workflows\n- Managing task delegation and parallel execution\n- Synthesizing results from multiple agents\n- Maintaining plan.yaml state\n\nTo get the best results, consider:\n- Start with the Orchestrator for any complex project\n- Provide clear goals and constraints\n- Review the plan.yaml before execution\n- Use the walkthrough summaries to track progress"
+    },
+    {
+      "path": "agents/gem-researcher.agent.md",
+      "kind": "agent",
+      "usage": "recommended\n\nThe Researcher gathers codebase context, identifies relevant files/patterns, and returns structured findings. It is typically invoked by the Orchestrator with a specific focus area.\n\nThis agent is ideal for:\n- Understanding codebase structure and patterns\n- Identifying relevant files for a specific feature\n- Gathering context before making changes\n- Researching technical dependencies\n\nTo get the best results, consider:\n- Specify a clear focus area or question\n- Provide context about what you're trying to achieve\n- Use multiple Researchers in parallel for different areas"
+    },
+    {
+      "path": "agents/gem-planner.agent.md",
+      "kind": "agent",
+      "usage": "recommended\n\nThe Planner creates DAG-based plans with pre-mortem analysis, presents for approval, and iterates on feedback. It synthesizes research findings into a structured plan.\n\nThis agent is ideal for:\n- Breaking down complex goals into atomic tasks\n- Creating task dependencies (DAG)\n- Running pre-mortem analysis to identify risks\n- Getting approval before execution\n\nTo get the best results, consider:\n- Provide clear research findings from the Researcher\n- Review the plan carefully before approving\n- Ask for iterations if the plan is not optimal\n- Use the plan_review tool for collaborative planning"
+    },
+    {
+      "path": "agents/gem-implementer.agent.md",
+      "kind": "agent",
+      "usage": "recommended\n\nThe Implementer executes TDD code changes, ensures verification, and maintains quality. It follows strict TDD discipline with verification commands.\n\nThis agent is ideal for:\n- Implementing features with TDD discipline\n- Writing tests first, then code\n- Ensuring verification commands pass\n- Maintaining code quality\n\nTo get the best results, consider:\n- Always provide verification commands\n- Follow TDD: red, green, refactor\n- Check get_errors after every edit\n- Keep changes minimal and focused"
+    },
+    {
+      "path": "agents/gem-chrome-tester.agent.md",
+      "kind": "agent",
+      "usage": "optional\n\nThe Chrome Tester automates browser testing and UI/UX validation via Chrome DevTools. It requires Chrome DevTools MCP server.\n\nThis agent is ideal for:\n- Automated browser testing\n- UI/UX validation\n- Capturing screenshots and snapshots\n- Testing web applications\n\nTo get the best results, consider:\n- Have Chrome DevTools MCP server installed\n- Provide clear test scenarios\n- Use snapshots for debugging\n- Test on different viewports"
+    },
+    {
+      "path": "agents/gem-devops.agent.md",
+      "kind": "agent",
+      "usage": "optional\n\nThe DevOps agent manages containers, CI/CD pipelines, and infrastructure deployment. It handles infrastructure as code and deployment automation.\n\nThis agent is ideal for:\n- Setting up CI/CD pipelines\n- Managing containers (Docker, Kubernetes)\n- Infrastructure deployment\n- DevOps automation\n\nTo get the best results, consider:\n- Provide clear infrastructure requirements\n- Use IaC best practices\n- Test pipelines locally\n- Document deployment processes"
+    },
+    {
+      "path": "agents/gem-reviewer.agent.md",
+      "kind": "agent",
+      "usage": "recommended\n\nThe Reviewer is a security gatekeeper for critical tasks. It applies OWASP scanning, secrets detection, and compliance verification.\n\nThis agent is ideal for:\n- Security code reviews\n- OWASP Top 10 scanning\n- Secrets and PII detection\n- Compliance verification\n\nTo get the best results, consider:\n- Use for all critical security changes\n- Review findings carefully\n- Address all security issues\n- Keep documentation updated"
+    },
+    {
+      "path": "agents/gem-documentation-writer.agent.md",
+      "kind": "agent",
+      "usage": "optional\n\nThe Documentation Writer generates technical docs, diagrams, and maintains code-documentation parity.\n\nThis agent is ideal for:\n- Generating technical documentation\n- Creating diagrams\n- Keeping docs in sync with code\n- API documentation\n\nTo get the best results, consider:\n- Provide clear context and requirements\n- Review generated docs for accuracy\n- Update docs with code changes\n- Use consistent documentation style"
+    }
+  ]
 }