docs: add DOCKER_MCP_IN_CONTAINER bypass information

[stellarshenson]

Summary

• Add documentation for DOCKER_MCP_IN_CONTAINER=1 environment variable bypass
• Helps users running in WSL2, containerized environments, or Docker CE
• Addresses "Docker Desktop is not running" error when Desktop backend sockets are unavailable

Changes

• Added GitHub-style NOTE infobox in Usage section of README.md
• Documented environment variable and its use cases

Context

Users running Docker Desktop in WSL2 or containerized environments may encounter
"Docker Desktop is not running" errors even when the Docker daemon is active. This
occurs because the Desktop backend API sockets (~/.docker/desktop/backend.sock)
are not available in these environments.

The DOCKER_MCP_IN_CONTAINER=1 environment variable allows users to bypass the
Desktop feature flag checks (see cmd/docker-mcp/commands/root.go:48), enabling
the MCP Gateway to work in these scenarios.

[slimslenderslacks]

Thank you @stellarshenson. I think we're also adding a new environment variable for use with Docker CE environment variables. Let me check.

[stellarshenson]

Super, this is one of the tools that we have included in our data science workbench - https://github.com/stellarshenson/stellars-jupyterlab-ds

[bobbyhouse]

I suppose this issue I ran into today when running Docker CE is related:
#293

Is there a way that we can infer this without having to explicitly set the env var?