Changes

testsuites: Add tunend-2trex test env

In order to validate tunend scenario (encaps/decaps) against an UPF, we
use TRex in stateful mode (ASTF) with its tunnels_topo infrastructure.
This setup is less performant than the stateless mode used in the tunmap
scenario, and as a result less throughput can be generated.

This really poses a problem if the host running TRex (generating the
GTPU traffic) is the same or less poerful hardware than the UIT running
the UPF, as generating+handling traffic is at least as complex as the
work UPF is doing.

Hence, in order to be able to use less performant hardware to generate
the traffic, a new test environment is added which expects 2 hosts
instead of 1 running TRex. As a result, the tunnels configured their
traffic are split into 2 segments/buckets of IP addresses/TEIDs, each
configured on the UPF by its own osmo-upf-load-gen over PFCP, and also
traffic for each of those two being generated/handled by one of the two
TRex instances being run.

In order to get the benchmark results, one must add the output from the
console from both TRex instances running in parallel on separate hosts.
With this it becomes possible to fill in the 100Gbps using C240-1 and
c240-2 hosts.

Change-Id: I6dcc3a27ef61857c05ab8d5cb67cceab5bb3dffa