Integrate webr and Pyodide to allow analysis of the samples

[WardBrian]

Similar to #57, but rather than before the sampling run, this would be done afterward.

Ideally this would look like a single script file and an output area that shows both stdout/stderr and plots. See https://www.cararthompson.com/posts/2024-02-01-experimenting-with-webr/ for some prior art on this.

A pre-populated magic variable called samples would make sense here. In terms of libraries to support, I think a nice target use-case is being able to use posterior + bayesplot in R, and arviz in Python.

[WardBrian]

Pyodide half: #120