Dev parser

[MichalPalic]

Hi, in combination with the one line change in fpgaconvnet-hls these are the small changes necessary to handle the unusual bias dimensions.

Note that since the last series of commits with the HLS and Chisel specific ConvolutionLayers, the ability to load conv layers has been broken due to issues with the inclusions in Network.py. I understand this is likely still under development but I'm including the stack trace for your convenience.