Adding gradient via convolution functionality to stage 4

[skrausse]

Integration of calculation of a gradient convolution in stage 4.
Steps done:

• Including of a new script gradient.py that takes some analogsignal, transforms it to an image sequence and performs a gradient convolution on that signal. Script is structured analogous to horn_schunk.py.
• Updated annotations in horn_schunk.py to make it identical to gradient.py. Kernel-type used was missing from the annotations and was therefore added.
• Including new rule in Snakefile of stage04 that can execute the gradient script (in analogy to optical_flow rule).
• Included needed parameters in the stage04 config template file.
• Resolved ambiguity in parameter naming convention between optical flow and gradient by including prefixes "FLOW" and "GRADIENT" respectively.