[New Sample] End-to-end Strands agent demo with AWS Agent Registry, ECS Fargate, and local MCP

## What this adds

A new advanced sample demonstrating a fully deployed, registry-driven Strands agent on ECS Fargate.
The agent has no hardcoded skills or tools — it discovers capabilities at runtime by querying the
AWS Agent Registry.

Proposed location: `[01-features/07-centralize-and-govern-your-ai-infrastructure/03-registry/03-advanced/registry-strands-ecs-mcp/`

## What makes it different from existing samples

This sample shows a production-style deployment where:

- A Strands agent running on ECS Fargate queries the registry per-request to find the right skill
- Skills are stored as `SKILL.md` files in `AGENT_SKILLS` records
- An MCP server (FastMCP, ECS Fargate) is registered via `synchronizationType="URL"` — the registry
crawler auto-populates tool schemas; the agent reads the MCP URL from the registry at startup
- Only the MCP tools declared in a skill's frontmatter are loaded per-request (selective loading)
- Full CloudFormation stack: VPC, ECS, internal ALBs, API Gateway + VPC Link, Cognito, IAM

## Components

- `deploy/infra/cfn.yaml` — CloudFormation stack (all infrastructure)
- `deploy/agent/` — Strands agent (FastAPI, Strands SDK, SigV4 MCP transport)
- `deploy/mcp/` — MCP server (FastMCP, financial data tools)
- `deploy/chat/` — Chat interface (FastAPI, SSE, React frontend)
- `deploy/infra/setup.py` — One-time registry setup via boto3 (no CloudFormation support for registry yet)
- `deploy/infra/register_skills.py` — Add new AGENT_SKILLS records to an existing registry
- `my_skills/` — Five financial analysis skills as SKILL.md files
- Jupyter notebook walking through the architecture and setup steps

## Domain

Financial analysis (5 skills: KPI calculator, cost efficiency, revenue growth, trend analysis,
executive briefing).

## Happy to adjust scope, location, or format based on maintainer feedback.

Start with the Jupyter notebook and the proposed sample directory, then read deploy/infra/cfn.yaml to understand the full stack. Review deploy/agent/, deploy/mcp/, deploy/chat/, setup.py, register_skills.py, and my_skills/ to map the components; done means the advanced sample deploys the registry-driven Strands agent, MCP server, chat interface, infrastructure, and five financial skills described in the issue.

Written by the indexing model from the issue text.