Split mrn2graph.json #7
Description
At the moment, we generate the entire index as a huge file that takes ages to load.
Instead, makemrn2graph.py could generate a folder of year+mtn prefix chunks (e.g.06JERUSALEM.json) and browser can keep them cached as a 2-level object.
In order to know what is a valid prefix (and avoid unnecessary http requests), the initial cache will be an object (also generated by makemrn2graph.py) mapping each existing prefix to null or somesuch. This should be way faster to load than the monstrosity we currently have.
GUI-wise, this would require temporarily turning the 2. select a graph section into a "loading" strip as long as there are pending mrn2graph fetches (could be more than one, since the search is multiline).