African scenarios and map fixes

Clean up and rerun the previously included African scenarios: 2003 Boumerdes (Algeria), 2004 Al Hoceima (Morocco), 2016 Bukoba (Tanzania), 1992 Cairo (Egypt)
Integrate content from the 2003 Marrakesh-Safi (Morocco) from the marrakesh branch that @raoanirudh was working on; add new data available from USGS (finite fault, stations)
Take 1989 Salima (Malawi) out of 'DRAFT' mode (unit test failure due to https://github.com/gem/earthquake-scenarios/issues/26)
Correct some broken code and maps in the src functions (e.g., addreses https://github.com/gem/earthquake-scenarios/issues/23)
Added 2009 Karonga (Malawi) (unit test failure due to https://github.com/gem/earthquake-scenarios/issues/26)

Some notes about the 2023 Marrakesh-Safi event, which likely could be improved.

USGS now has 3 seismic stations (see here). One of these stations has a PGA of ~0.79g, which can significantly increase out estimated ground shaking when we condition to the seismic stations.
Previously, a point rupture from USGS was used, but now there is a finite fault estimate. Should we consider keeping both for reference? This version only includes the finite fault estimate and not the point rupture.
In Slack, @raoanirudh shared some fatality estimates by Adm2, but I didn't add those since I didn't know the source
I've seen some presentations that included impact at the subnational level, but would need to reach out to the presenters to find the source

gem / earthquake-scenarios