Planned improvement in the algorithm:

1. try to match the segments via MID (like currently implemented, with source compare check)
2. remove the segmentNrInTask search
3. Add for each segment a segment hash (based on the segment source compare function used currently in relais mapping)
   1. Best would be to create the hashes per file, since only one file is compared with another. This reduces memory usage, when there are packed a lot of files in one ZIP container.
4. use the diff algorithm to compare the hash map of the source file and the relais file.

1. Use that output to merge the segments.