Methodology tweaks for modal metrics

[admrtin]

As I was working on #165 , I noticed some minor things that needed to be adjusted about how we are computing the modal metrics (WCI and LTS).

• Remove the nearest neighbor search in walk eligibility. Walk eligibility should be determined only by the way of interest's attributes, not neighboring way's attributes.
  • This caused inflation of WCI values for ways that were inherently unwalkable such as motorways, and vice-versa for ways that are very walkable.
• way_is_sidewalk and way_is_footway should have highway tag matches before way_is_footway.
• In addition to the edge's cycleway=? tag, the CyclewayTag object should also query the edge's highway tag (if highway=cycleway, return Cycleway::DedicatedWithBuffer)
• The MIN_DISTANCE_RTREE_NEIGHBOR constant in .../common/mod.rs should be decreased to a decimal degree value that correctly represents 15m distance.
  • This value was previously set to 0.0001378 degrees, which is ~15m , but since this value is squared in the RTree neighbor search neighbors are found in a 1.3km radius (~87x off of 15m). We should square this distance, which results in ~1.8e-8 degrees.
  • This also means that our performance will increase drastically since we no longer have 1.3 km of neighbors to weigh 🚀.
• Add Highway::Cycleway as a walkable highway tag.
• As a part of this issue, in bulk_compute_modal_metric, we should change the CLI input metric_name (string) to modal_metric (enumeration)
• Add pedestrian/ living street check for Max WCI