[parallelization] Correct boundaries for uv-map in grid-structure

[MartinPyka]

The UV-map of DG is placed somewhere in the middle of the whole uv-area. This seems to be the reason, why only one half of the post-synaptic neurosn are connected with EC2-neurons. Maybe the boundaries of the grid-structure of the synaptic layer are determined in the wrong way?

[pherbers]

Fixed in 17c8458.
The problem was in _uv_to_cell_index() and _cell_index_to_uv(), which could not handle different resolutions for rows and columns, causing the kernel to be stretched.
I'm not sure how heavily the other functions rely on this and if anything else has changed drastically. Is the _resolution field still used or can it be replaced completely by _row and _col?

[pherbers]

The grid structure should work correctly now. The uv boundaries are used correctly to make use of the grid space more efficiently. Mappings now look like those in master, but please confirm if they are correct.
Grid resolution is still hard coded in pam.py:1044 computeConnectivity(), should this be an option in the UI?