/RundownMaxMB should rely on ETW MaximumFileSize

[suprak]

We see cases where something kills PerfView during rundown phase, resulting in a runaway rundown.etl reaching 100+GB and flooding disks.

Turns out the /RundownMaxMB flag is implemented using logic that watches the file size while PerfView is running.
However if something happens to the PerfView process, then there is nothing left watching the rundown.

So implementing the max file size using a native ETW mechanism would be best.