The form discussion: https://forum.camunda.org/t/how-to-listen-for-a-link-intermediatethrowevent-link-start-event/1187/3

As described in the forum post it would make sense registering the throwing link events in the history in order to find out which event triggered a catch event in case of multiple throwing events.

This "bug" or "conception difference" has been encountered while developing the process test coverage library: https://github.com/camunda/camunda-process-test-coverage/tree/master-7.5.0

The coverage can't reach 100% unless there is a way to listen for throwing link events (or unless a workaround is in place).