Dockerise MCP-run-python server + add git workflows to publish images

[stdweird]

Fixes #21

Shamelessly taken from pydantic/pydantic-ai#2457

[stdweird]

I have taken the verbatim code from the pydantic-ai PR, but i see e.g. that the latest denoland container at time of writing is already at 2.5.6
not sure if it should be updated or not?

[stdweird]

I also don't know how to credit the original author. @assadyousuf feel free to open your own PR, then i'll close this one

[stdweird]

Not sure what i was thinking to open this PR before understanding what is going on ...

Anyway, i think it makes more sense to start from a python base container, and run uv sync or somethign and have uv mcp-run-python as an entrypoint so it's easier to pass mcp-run-python args instead of deno args

[stdweird]

This gives a container that you pass the options to mcp-run-python with (podman is like docker, and the hash is the container id)

$ podman run 75849b9f8e02 --port 5001 --deps numpy,scipy streamable-http
Installing dependencies ['numpy', 'scipy']...
Running mcp-run-python via streamable_http on port 5001...
Listening on port 5001

[stdweird]

@DouweM i took the github actions from the other PR