We shouldn't need TicketHolder to have the complex virtual interface it does today after the linked ticket to remove PriorityTickeTHolder is complete. At a minimum we should be able to remove the indirection around the queueing stats and consolidate the code implemented at the TicketHolder and SemaphoreTIcketHolder layers today.

One obstacle to consider is the MockTicketHolder used in the execution control unittests today. I suspect we can provide what this test needs without inheritance to fully simplify this interface into a single class/implementation, but if not we should still make it as minimal as possible.