Starter project: Optimize map importer

[dabreegster]

Importing all of London, https://github.com/dabreegster/abstreet/blob/7a0dbb28b1e711bb9a9f7265ff95e3ee411089df/map_model/src/make/parking_lots.rs#L141 took 40 seconds. It's easy to use timer.parallelize to speed this up. Running the importer on large areas, checking the logs for the slowest steps, and applying simple fixes like parallelize and FindClosest is a great way to learn a bit about the map importing process.

[at-tran]

Is this issue solved? If not I would like to take a crack at it.

[dabreegster]

@at-tran: I just reimported the largest maps, and the time is now almost totally spent building the contraction hierarchies for pathfinding. All of the other inefficiencies have been fixed, so I'll close out this issue.

If you're interested in data structures, #54 is another optimization task that's a ~gentle introduction to the simulation layer. Or if you're particularly interested in map importing, simulation, graphics, etc, I could point you towards other good starter bugs.

[dabreegster]

#85 may also be of interest if you have knowledge of downsampling timeseries data

[at-tran]

Thanks for the suggestions! I'll take a look at those issues and decide what I want to work on