- changed data_to_images in order to use the KM3Pipe Pipeline concept instead of iterating over the HDF5Pump.
- Added output dirpath, chunksize, compression filter and compression strength arguments
- Added CI versioning and publishing of OrcaSong via Pypi

- Add folder structure to config files folder

Add 'tight_0' timecut, which is useful for ORCA 115l mupage vs random_noise vs neutrinos classification.

- Implemented a better parser and implemented config files as an input to OrcaSong.
- Got rid of the ORCA_Geo_115lines.txt file, the detector geometry is now read from a .detx file that needs to be supplied with the input file.
- Improved general readability of the code
- fixed run_id readout
- added option to make no timecut

First prototype of docs included!

- Change folder structure for continuous integration and to better suit standard python package structures
- add Dockerfile, Jenkinsfile, requirements for CI
- add License, Changelog
- remove globals file, not necessary anymore
- simplify geo binning, change geo_fix parameter to det_geo parameter
- better readout of run_id variable

E.g. if you use the 1-5GeV & the 3-100GeV ORCA 2016 MC, it is not possible to distinguish the >3GeV events from the lowE prod from the events of the highE prod.

2) Now, always a fixed random seed is used. This is needed, if you use multiple input datasets and you want to throw away a fixed portion of events.

- Add possibility to throw away a certain percentage of events in a file

- Corrected run_id readout due to new data structure
- Added time_residual_vertex to readout
- Changed event track datatype from float32 to float64 for better precision with times