Pyodide notebooks support

[danyx23]

This PR is only a proof of concept at this stage - it demonstrates using pandas inside pyodide based notebooks on grapher pages - i.e. python and the pandas/numpy stack run entirely in the browser, without any server component. Data is fetched via our file based API just like grapher (the data is fetched again but should be fast due to client side caching).

To play with it, open the staging server at a /grapher/ url and scroll to the bottom.

There are a lot of things that are not ideal about this proof of concept, e.g.:

• The "library code" that is taken from Lars' recent new pythonic api to access data for grapher charts is copied verbatim into the first cell instead of used from a library
• The notebook infrastructure we use, starboard, has not been developed further since early 2023 or so. It is thus stuck at pyodide version 0.20. The current version, 0.25, comes with a bunch of nice additions, e.g. transparent support for the requests library
• If we had access to 0.25 then Lars' pythonic api code could be used verbatim - as is, http requests had to be replaced in the code with pyodide.http.open_url
• The notebook text is hardcoded in the GrapherPage.tsx file

[owidbot]

Quick links (staging server):

Site	Admin	Wizard

Login: ssh owid@staging-site-wasm-notebooks

• Site-screenshots: https://github.com/owid/site-screenshots/compare/wasm-notebooks
• SVG tester: https://github.com/owid/owid-grapher-svgs/compare/wasm-notebooks

SVG tester:

Number of differences (default views): 424 (faf803)
Number of differences (all views): 890 (47426c)

Edited: 2024-05-29 16:05:00 UTC
Execution time: 1.24 seconds