IPv6: examples: fix degenerate traffic parameters in nclients

The Telnet traffic in examples/ipv6/nclients was parameterized so
sparsely that the statistical test effectively measured RNG noise
rather than behavior. With a single client (n=1),
numCommands=int(exponential(1)) (which is 0 about 63% of the time) and
commandLength=exponential(1B), a whole run carried only a handful of
mostly-control packets. Any RNG-stream shift between commits flipped
entire sessions in and out of existence, so the test failed spuriously
and could not distinguish a real regression from a reseeding.

Align the traffic with the sibling examples/inet/nclients example:
n=4 clients, numCommands=int(exponential(10)), commandLength=
intWithUnit(exponential(10B)). Router-forwarded packet counts rise from
~20-40 to ~2600, making the statistics meaningful and robust.

Keep startTime=uniform(10s,15s) rather than the IPv4 example's
exponential(5s): IPv6 needs ~2s of ND/DAD before a host can connect,
and clients that start earlier never establish a working session
(observed as 0-packet clients under exponential(5s)).

Author: avarga

examples/ipv6/nclients/omnetpp.ini

@@ -10,20 +10,21 @@
 abstract = true
 
 # number of client computers
-*.n = 1
+*.n = 4
 
 [Config TCP_APP]
 abstract = true
 
 # tcp apps
 **.cli[*].numApps = 1
 **.cli[*].app[*].typename = "TelnetApp"
+**.cli[*].app[0].localAddress = ""
 **.cli[*].app[0].localPort = -1
 **.cli[*].app[0].connectAddress = "srv"
 **.cli[*].app[0].connectPort = 1000 #same destination port numbers
 **.cli[*].app[0].startTime = uniform(10s,15s)
-**.cli[*].app[0].numCommands = int(exponential(1))
-**.cli[*].app[0].commandLength = intWithUnit(exponential(1B))
+**.cli[*].app[0].numCommands = int(exponential(10))
+**.cli[*].app[0].commandLength = intWithUnit(exponential(10B))
 **.cli[*].app[0].keyPressDelay = exponential(0.1s)
 **.cli[*].app[0].commandOutputLength = intWithUnit(exponential(40B))
 **.cli[*].app[0].thinkTime = truncnormal(2s,3s)