Fix guidance sliproads scenario 3

[oxidase]

Issue

PR fixes scenario 3 of #4348

by adding an additional check for an intermediate node

Also PR changes allowed angle from 115 to 140 degrees for non-link sliproad check that updates scenario 2 of #4348

Tasklist

• add regression / cucumber cases (see docs/testing.md)
• review
• adjust for comments

Requirements / Relations

Link any requirements here. Other pull requests this PR is based on?

[daniel-j-h]

Can you make maps for before / after, e.g. for Berlin and SF. I would love to understand what will change here with this addition. I think you can use the geojson logger we have for this use-case.

Also do you think we should always only skip a single intersection? Or should we skip intersections until a distance threshold is reached?

[daniel-j-h]

Travis fails: dk:rural seems not to be in taginfo.json.

I don't know why we didn't see this before.

[oxidase]

@daniel-j-h i have created a map where

• read nodes are no more sliproads because of a new double announcements check (cases like http://map.project-osrm.org/?z=18&center=39.509808%2C-121.539978&loc=39.510303%2C-121.539626&loc=39.509653%2C-121.540584&hl=en&alt=0 or http://map.project-osrm.org/?z=18&center=32.689455%2C-117.171555&loc=32.689295%2C-117.170949&loc=32.690094%2C-117.171421&hl=en&alt=0)

• blue nodes are new sliproads because of the change angle from 115 to 140 degrees

• green nodes are new sliproads with an intersection in a candidate road

As i see blue nodes is really big change, so i think to remove

osrm-backend/src/extractor/guidance/sliproad_handler.cpp

Line 436 in ad9ef92

if (deviation_from_straight > 180. - minimal_crossroad_angle_of_intersection &&

condition but it will require adding _link in OSM sliproad ways to keep some "green" sliproad nodes.

[daniel-j-h]

Just sampled some locations affected by this change and it looks good from what I've seen so far. 👍