Performed a test. Detailed steps to reproduce the issue:

1. Import *.bpmn2 process attached to business-central (BPMS or RHBA);
2. It will require a "brmsuser" set since there is an user task assigned to it (or just set another Actor for "SampleHT" existing User task placed inside subprocess);
3. after build&deploy, start the process definition - it should get stopped in "SampleHT" user task;
4. now send a signal "name: signalTest - value: signalTest" to this process instance;

Result: At this point, the process instance is waiting an action in SampleHT user task. However, if you send a signal the catch signal event will fire anyway. So if you look now at nodeinstancelog table, it will have a register for that signal.

5. Then, start and complete SampleHT User Task and send signal event to process instance again.

Result: It is now waiting or signal get fired or InnerHT user task to be completed. Therefore, after sending a new signal... the catch signal event will be fired and process terminated.