Support median

[mccalluc]

• Fix Support medians #323
• Ignore a non-deterministic matplotlib warning
• Factor out shared analysis shiny inputs: Besides shortening the code, this reduces the chances of IDs getting out of sync.
• Up the minimum contributions to 2 temporarily to avoid calculating median: contributions=1 + public_info="lengths" causes unable to infer bounds opendp#2331 -- but the real fix is With delta, margins can be dropped #262
• Add an option for downloading the python source that is used to generate the notebook. Probably only useful for developers, but not obtrusive.
• Add another column, self_assessment, to the demo CSV: With more analysis options, it is useful to have more data to analyze.
• In the generated notebooks, names were colliding. Change the binned data from column_config to column_bin_config.
• Add markdown section headers to generated notebook to improve readability.

For the moment, it just makes 100 candidates between the lower and upper bounds. This is probably be changed, but I'm not sure what the right design is. Opened an issue, but shouldn't try to resolve that in this PR.

• Best prompt for quantile candidates? #304
• see also: Polars: Quantile candidates could just take bounds opendp#1706

Reporting details are still TODO:

• Fill in the visualization/report/etc. for means and medians #275

[mccalluc]

Filed an issue upstream:

• calculating median: contributions=1 + public_info="lengths" causes unable to infer bounds opendp#2331

Will wait for feedback on that before proceeding here.