Switch to scapy with python 3.

Do not use privileged or NFS source ports for UDP packets as inetd
ignores such packets. This should avoid some sporadic failures.
While there, use variable names consistently in all tests.

By removing all the sleeps I created a race in the pf tests. The
packet was sent and its reply received before the sniffer was up
and running. So sleep a second after starting the sniffer thread
bu

By removing all the sleeps I created a race in the pf tests. The
packet was sent and its reply received before the sniffer was up
and running. So sleep a second after starting the sniffer thread
but before sending the packet. With pf_forward 0m51.54s, pf_fragment
1m33.59s, pf_state 0m03.64s the test are still faster than with
forking.

show more ...

Replace fork() and sleep() with a Python thread for sniffing packets.
This reduces test execution time from 2m21.95s to 1m09.80s.

Fix tests: Restrict getpid() to lower 16 bit so that it can still
be used as packet id. Now scapy calls nexthopmtu with this name
explicitly in icmp structure.

Use a simpler expression to check the ether type in scapy. This
makes the fragment tests work on FreeBSD. From Ilya Bakulin.

Kill trailing whitespaces.

Add tests for pf fragment reassembly with all kinds of overlapping.
Some fragments contain invalid data, this must be replaced to
reassemble a valid packet.