Thanks for posting the links.
One thing I learned in bicycle crash analyses is to look beyond the standard crash types (which are useful to some degree) and try to assess any specific behavior patterns that might be a factor in crashes, especially at complicated intersections.
Vancouver has a remarkably low number of bicyclist crashes for its size, including fatalities. Many factors could be affecting this, but at least it's a good trend.
There are a lot of doorings, though. Dooring crashes are often common in dense urban environments with on-street parking. One way to completely avoid dooring is to ride outside the door zone, which is taught by several "smart cycling" programs. There may be factors that work against such behavior, such as motor traffic speed in the lane adjacent to parking, or conversely congestion in that lane where motor traffic is slower than cyclists (e.g. cyclists trying to get through by riding in the space between slower and parked vehicles & getting doored).
The countermeasures list contains many recommendations for increased use of traffic control devices, such as signs and markings. Historically in North American practice, passive devices that try to change or contradict ingrained road user behavior at intersections have shown limited effectiveness in changing behavior - this could be a factor in some of the turning-movement crashes at designated bikeways. Active devices such as signals can have increased effectiveness, but this depends on road user compliance. I remember observing a bicycle signal in Montreal several years ago with almost zero cyclist compliance, but due to low motor vehicle travel speeds there were many conflicts and people yelling at each other in French, but few reported collisions. If motor traffic speeds are faster at these locations, there may be a higher crash rate.