Distances calculated with bike_distmat() seem incorrect about bikedata HOT 1 CLOSED

Thanks @JorisKlingen! This was a very important bug that you found, caused by an overseen mismatch between 0-based C++ indices and 1-based R indices in the dodgr package used to calculate these distances. With this commit, your example now yields only 139 from > 500,000 station pairs for which great circle distances are greater than street network shortest paths. These arise in situations where the station coordinates map onto parts of the street network that are closer together in network terms than the great circle line between the station pair itself. They are only for station pairs that are very close together, and are appropriate in these cases, reflecting as they do slight mismatches between given station locations and the underlying OSM street network, and for which the distance mis-matches are very, very small.

from bikedata.