Research Request - Second try for low speeds + fix only 1 vp within segment

[tiffanychu90]

Complete the below when receiving a research request, and continue to add to this issue as you receive additional details and produce deliverables. Be sure to also add the appropriate project-level label to this issue (eg gtfs-rt, DLA).

Research Question

Single sentence description: Add a step after identifying unusually low speeds on segments to select another vp and try again. While we're doing this, handle the fact that there are segments with only 1 vp.

• Epic - Daskify RT segment speeds #592

Detailed description: Do the linear referencing, and before saving out the speeds by trip, do an extra check to find these low speeds and try again to calculate.

• From July numbers, it looks like about 25% of segments do not pass the threshold of covering at least 30% of segments. Of these, if we try and recalculate again, there's still another 20% that are still erroneous (this is 0.25*0.02 = 0.06 of all segments).
• This try again should also handle the fact that there are segments with only 1 vp. Select the vp right before, even if it's not within the segment. If we still can't, select the vp right after.
• The second try: should it be done against the whole shape_geometry (projecting it is tricky because shape_meters will be inconsistent using a complex shape), or should it use a straight line distance (which ignores the shape path completely)?