Reason for version 0.3: config-file based

+ rework the paths (and make them absolute)
      --> let the app know where to place the output dir
      --> the data + output dir now depend on the location of the config file
+ implement options for automatic out-dir naming
+ restore parallel-filter option

... to allow modify them later on

(For status updates when no carriage return desired)

--> less memory & less computation time.
numpy's FFT would always work in double precision

(only used in post-precessing for determining max. amplitudes)

by constructing a custom taperd window (more smooth at nyquist than the previously used tapered cosine)

----
Also: one implementation to remove the Gibbs Phenomenon. But will not be useful here since the signal will be padded with zeros --> not sufficient --> commented out
Source: http://ieeexplore.ieee.org/document/902128

works really great!

Sources of inspiration:
https://www.dsprelated.com/showthread/comp.dsp/15995-1.php
http://ieeexplore.ieee.org/document/1450934/

Since we are in the frequency domain anyway, make direct use of it's advantages (and speed up)

--> new config setting: config.FFT_filt

e.g.:
 - optimizing the interval of the adaptive-interval reate decay plot - to include remaining events in last bin
 - various fixes
 - beautifications

GMT scripts: new theme: Basel with pressure data, but no histogram + fixes
SQL script: output publicid as eventid in first column

make it identical to non-mpi version