initial tests for docker setup

[Shashankss1205]

Reference Issues/PRs

Closes #444 . Adds Docker containerization support for sktime-mcp.

What does this implement/fix? Explain your changes.

Adds end-to-end Docker support for running sktime-mcp without any local Python setup. This includes:

New files:

• Dockerfile - Multi-stage build (builder → slim runtime). The builder stage installs build tools (build-essential, gfortran, libopenblas-dev) and all Python dependencies, then only the installed packages are copied into a minimal python:3.12-slim runtime image. Runs as a non-root mcp user for security.
• .dockerignore - Filters out .git, __pycache__, .venv, docs/_build, and other dev artifacts to keep the build context small (~456KB).
• docker-compose.yml - Service definition with configurable environment variables (SKTIME_MCP_LOG_LEVEL, SKTIME_MCP_AUTO_FORMAT, etc.) and a named sktime-models volume for persisting saved models across container restarts.

Modified files:

• README.md - Added a 🐳 Docker section under Installation with build/run/compose commands and a Claude Desktop JSON config example for Docker-based MCP. Updated the project structure tree to include the new files.
• Makefile - Added make docker-build and make docker-run convenience targets.

Does your contribution introduce a new dependency? If yes, which one?

No new Python dependencies. Docker is the only external requirement, and it is optional - existing pip-based installation is unaffected.

What should a reviewer concentrate their feedback on?

• The multi-stage Dockerfile approach - builder installs build-essential/gfortran/libopenblas-dev for compiling numpy/scipy, but only libopenblas0 is carried into the runtime image. Is this the right trade-off vs. a single-stage build?
• The docker-compose.yml uses stdin_open: true with tty: false - this is intentional since MCP JSON-RPC needs raw stdin, not a pseudo-TTY. Worth confirming this works across Docker versions.
• Claude Desktop config in README uses "command": "docker" with "args": ["run", "-i", "--rm", "sktime-mcp"] - reviewers familiar with MCP client setups should verify this pattern.

Any other comments?

Verified end-to-end locally:

• ✅ docker build completes successfully (~81s)
• ✅ All package imports work inside the container (sktime-mcp v0.1.0, sktime, mcp)
• ✅ MCP initialize JSON-RPC handshake succeeds
• ✅ tools/list returns all 22 registered tools

PR checklist

For all contributions

• I've added unit tests and made sure they pass locally (make check).
• I've added the tool to the online documentation in docs/source/.
• I've updated the existing example scripts or provided a new one to showcase how my tool works in examples/.