Improve codebase in the wake of MapLibre-GL migration

[jjnesbitt]

There are several things that remain less than ideal immediately following the merging of #63, which should be addressed.

1. The complexity and multi-purpose role of updateVisibleMapLayers
2. The handling of datasets with multiple networks
3. The handling of network datasets when populating simulation results
4. The handling of datasets in `toggleNodeHandler'
5. The default value for circle layer opacity

There are further things that could be addressed at a more architectural level, but these were the immediate things to fall out of the maplibre implementation.