* not done. The 'setsockopt(SO_RECVBUF)' system call cannot set the socket buffer
size lager than the specified in net.core.rmem_max. The requested value
was 2 GB (commit 3b0e2e6), which is far too large for this application,
so it was restored to the acceptable 100 MB value.
* The syscall does not fail if the requested buffer size is larger than
net.core.rmem_max. Use 'setsockopt(SO_RECVBUFFORCE)' to actually force a
value larger than the system limit, which can be done if run in a
privileged context (capability CAP_NET_ADMIN set).
* The real value is read with 'getsockopt(SO_RECVBUF)'. If it
corresponds to twice the requested value (see 'man 7 socket'), it is
printed in green, otherwise it is signalled in red.
* The 'setsockopt(SO_RECVBUFFORCE)' syscall removes the need to write to
/proc/sys/net/core/rmem_max, so this was was suppressed in the
'UDPStandardImplementation' constructor.
* The test on EIGER detectors before setting the system
buffers was removed. Was there for 9m/2m eiger, but one can take care of
memory requirements using a customizable max socket buffer size(only with
permissions). to be implmented later.
* The file /proc/sys/net/core/netdev_max_backlog is first read by the
receiver to check is the current value is OK. If it is not, the receiver
directly writes the good value into the file (instead of delegating to
the system shell), printing a red error message if there is an access
error (non-privileged user).
