Document how service names are chosen in the trace summary screen

[codefromthecrypt]

We have unit tests, etc, and generally most zipkin instrumentation logs zipkin annotations.

However, we occasionally get questions by folks who aren't logging zipkin annotations, or are otherwise curious about why a service name is chosen. By documenting the algorithm, we can help them find bugs or constraints that could lead to a better understanding or experience.

https://github.com/openzipkin/zipkin/issues/962