No flush after event number output #150

Hi Oleksii,

I am not sure if this is a real issue and if there are some reasons for why it is implemented this way, but let me explain first:
I am using the feature of the TaskManager to output the current event number at a fixed frequency (https://github.com/HeavyIonAnalysis/AnalysisTree/blob/master/infra/TaskManager.cpp#L118 ). When running locally in a terminal this works fine, however, when trying to write this into a log file when running on virgo the buffering makes it a bit useless. Since there is no "flush" or "std::endl" at the end of std::cout << "Event no " << iEvent << "\n"; the output is not directly written into the log file but buffered until the buffer is full or something else triggers a flush and then its written all at once.
In my case for example when I run over 20k events with a verbosity period of 500, I cannot check how far the job has progressed because there is not enough std output to fill the buffer during the run, so all the "event no XXXXX" are written into the log at once when the program has finished.
I know that we want to minimize the number of write operations on the cluster, but it also does not really make sense to buffer the output which is supposed to track the current progress, right?

Thanks in advance
Adrian