[Battlemesh] The 'blowing up the network' results

[Juliusz Chroboczek]

An interesting data point is that Babel did reasonably well on the test
with marking, but horribly badly on the BE test.  I would guess that since
Babel was running with radio diversity enabled, it sent all the traffic
along a non-interfering route, while the other protocols were running pure
ETX/ETT, and therefore spread out the traffic more evenly.

Four-radio tests would be informative.

Looking forward to analysing the data in more detail.  Folks, this is
being actually productive.

-- Juliusz