Perfetto: Fixed a crash when the Mojo binding got torn down before the interfaceptr

The ProducerHost would reset the producer_endpoint_ member on connection errors,
but as these are connection errors on the InterfacePtr (i.e. outgoing calls) rather
than the Mojo binding (owned by a StrongBindingSet in the PerfettoService), additional
calls could still happen and would crash when accessing the already-reset producer_endpoint_.

The only reason to reset this member is as a safeguard against edgecases where callbacks
come from Perfetto while the ProducerHost is in a half-destructed state; we can accomplish
the same by just manually resetting the producer_endpoint_ in the destructor.

R=ssid@chromium.org

Bug: 945982, 947449
Change-Id: I44c54f4409ab384243b89ab161b171477849849b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1545082
Auto-Submit: oysteine <oysteine@chromium.org>
Reviewed-by: ssid <ssid@chromium.org>
Commit-Queue: oysteine <oysteine@chromium.org>
Cr-Commit-Position: refs/heads/master@{#646032}