Hqta collinearity

[edasmalchi]

resolves #1292 and greatly increases HQTA accuracy by ensuring we only aggregate frequency across routes where those routes are collinear. Use those results to filter down segments, reducing the chance of infrequent route segments getting matched to nearby frequent stops that aren't actually on that route, which creates a cleaner dataset with, importantly, far fewer erroneous major stops.

Also, don't consider looping segments for intersections, and ensure process for generating hq corridors and intersection-based major stops is filtered to bus only (rail major stops already captured in separate process)

see https://github.com/cal-itp/data-analyses/blob/e8b3c7f55bd5ba8dcf6838d1b4c086689475297a/high_quality_transit_areas/technical_notes.md for details on collinearity check

HQ Corridor Examples (old is blue):

New Major Stops:

[github-actions]

nbviewer URLs for impacted notebooks:

• high_quality_transit_areas/08_2024_methodology.ipynb
• high_quality_transit_areas/09_enforce_collinearity.ipynb