dhtperformance: total time display wrong when doing parallel requests

[gavin-norman-sociomantic]

I assume this happens with all request types, but I noticed it while doing Get requests:
./dhtperformance -S new-1-AP-dhtmemory.xml -m get -c usermap_10088180265682214291 -i 0 -n 1000000 -p 50

The total time listed on the console (avg: 198.99μs, count: 996600, total: 198.72s) increases at a rate much faster than reality. Either the process is running on an incredibly dense planet or there's a bug. (I suspect the timer is updated once for each parallel connection where it should only be updated once.)