Investigate and fix potencial race conditions on port stats multipart reply

[italovalcy]

Similar to issue #26, we should further investigate and eventually fix potential race conditions on port stats multipart reply.

As @viniarck comment on PR #28:

@italovalcy looks like the multipart handler of port stats also has a similar race condition vulnerability, should we map this in another issue to be fixed in another opportunity?

The tests I've been running show that the port stats message usually doesn't get fragmented at the OpenVSwitch nor Noviflow switches. However, I think we should map this to be handled in the future.