initial_time lost in the past with high packet loss

[taqen]

setting initial time before the network is ready, in the example scripts, leads to a situation where nodes receive wrong allocations for the allocation generator.

With high packet loss, the network join could take minutes, in this case after the network is ready, time would have been accumulated to the point where time() - initial_time is much bigger than the dataset, thus nodes receive nothing, and the logger logs nothing. The result of which, the network appears to perform better with higher losses.