Rember Intersection Shapes

Changes the processing order in the edge based graph factory.
Instead of iterating over all outgoing edges in order, we compute the edge
expanded graph in the order of intersections.
This allows to remember intersection shapes and re-use them for all possible ingoing edges.

Also: use low accuracry mode for intersections degree 2 intersections

We can use lower accuracy here, since the `bearing`
after the turn is not as relevant for off-route detection.
Getting lost is near impossible here.

include/extractor/guidance/coordinate_extractor.hpp

@@ -38,7 +38,8 @@ class CoordinateExtractor
                                             const NodeID to_node,
                                             const std::uint8_t number_of_in_lanes) const;
 
-    // same as above, only with precomputed coordinate vector (move it in)
+    // Given a set of precomputed coordinates, select the representative coordinate along the road
+    // that best describes the turn
     OSRM_ATTR_WARN_UNUSED
     util::Coordinate
     ExtractRepresentativeCoordinate(const NodeID intersection_node,
@@ -240,6 +241,14 @@ class CoordinateExtractor
                         const double segment_length,
                         const std::vector<double> &segment_distances,
                         const std::uint8_t considered_lanes) const;
+
+    // find the coordinate at a specific location in the vector
+    util::Coordinate ExtractCoordinateAtLength(const double distance,
+                                               const std::vector<util::Coordinate> &coordinates,
+                                               const std::vector<double> &length_cache) const;
+    util::Coordinate
+    ExtractCoordinateAtLength(const double distance,
+                              const std::vector<util::Coordinate> &coordinates) const;
 };
 
 } // namespace guidance