Feature Request: Add a count for total number of instruction for filtered trace

Created by: vicwlee

When traced in filtered mode, the total number of records does not equal to the number of instructions executed during tracing. It would be very useful to have accurate instruction count for filtered trace. At the minimum, it should provide a total number of instructions executed during the tracing period. Additional details such as # of instructions per thread, or # of instructions since last record (i.e. for every filtered miss) would great with the understanding that these will add additional overhead in speed and size.

This feature would be extremely useful to try to correlate filtered trace with unfiltered trace and would enable filtered trace be more useful for cache studies.