Path from *_rvalid_i to performance counter

Created by: Silabs-ArjanB

Combinatorial paths exist from the *_rvalid_i inputs to mhpmcounter_q[3]. The rvalid signals are typically late arriving inputs the the stated performance counter uses a 64-bit incrementer.

Would it be okay to count all events 1 cycle later (thereby preventing the performance counters from contributing to critical paths)?