Timeout in kernel-based ping-pong test

[pschichtel]

6a71495 seemed to have entirely fixed the lock-up issues in the ping-pong test both in the emulated and on the real socket. on the emulated socket it is still not reproducable, but for the the real socket it just got harder to hit.

The way the previous lock-up was fixed lets me believe this issue might actually not be related. This issue also seems to behave more consistently, as it usually locks up late into the test after at least 2400 messages. It also doesn't cause a send-timeout, so it probably loses/reorders a frame late in the message, possibly an exhausted receive buffer or an unfortunate GC event.

I'd estimate the probability with about 10%.

[pschichtel]

the userspace ISOTP implementation has been replaced by the ISOTP kernel module and a custom selector