Use better palette than viridis and/or annotations for non-sequential data (and sequential data when small differences matter)

E.g., consider qualitative palettes referred to in overviews [here](https://davidmathlogic.com/colorblind/#%23D81B60-%231E88E5-%23FFC107-%23004D40) or [here](https://www.nceas.ucsb.edu/sites/default/files/2022-06/Colorblind%20Safe%20Color%20Schemes.pdf), or maybe just Paired or something. For interactive plots, we may want to have a re-randomize palette option like in EpiVis.

Also consider adding annotations with `{ggrepel}` or manually with `geom_text` rather than relying on colors. Perhaps interactive plots makes this concern moot (unless forced to be non-interactive, e.g., when preparing manuscripts).

Start by locating the plotting entry points for the interactive and manuscript/non-interactive plots; the issue names qualitative palettes, ggrepel, geom_text, and EpiVis as relevant references. Define the intended palette and annotation behavior before changing anything, then verify that non-sequential and small-difference sequential data remain distinguishable without relying only on color.

Written by the indexing model from the issue text.