One common route to MSE fuzzer timeout is a fuzzed trun containing a large sample_count, all with sample_size=0 (typically via trex default_sample_size == 0 and trun tr_flags' sample-size-present flag not set).

We could arbitrarily cap or reject such sequences in our parser, though I'm averse to doing that unless usage data indicates such sequences are rare in real content for our users.

This bug tracks adding instrumentation to determine if such capping/rejection wouldn't impair actual user experience (while letting the fuzzers operate more efficiently).