Change README example to show clearer real-time bias

Currently the quickstart example uses data from the full outbreak, which will have limited bias (because most outcomes known).

Suggestion: filter data to focus on the early outbreak period. Also check that this isn't an issue that needs updating in vignettes too.

Would also be useful to point to code to reproduce the plot in the README, for interested readers.

Start with the README quickstart example and identify where it selects data from the full outbreak. Check the related vignettes for the same example and trace how the plot is generated. Done means the example focuses on the early outbreak period, any duplicated vignette content is consistent, and readers can find the reproduction code.

Written by the indexing model from the issue text.