Compilation error with concatenate layer

Created by: rdhen3

Hi there, I have just started using hls4ml a couple of days ago with the hope of applying it to a simple SSD inspired CNN. The model involves some number >= 0 of convolutional feature extraction layers well as several parallel sets of predictor layers whose outputs needs to be concatenated together.

A simple version of the model has the following keras summary:

The current stable build seems to have issues in processing concatenate layers in that i think the shape is incorrectly inferred leading to an IndexError in hls4ml/model/hls_layers.py:

   1380         axis = self.attributes['axis']
   1381         shape = inp1.shape[:]
-> 1382         shape[axis] += inp2.shape[axis]
   1383         rank = len(shape)
   1384         if rank > 1:

IndexError: list index out of range

I tried pulling the latest master upon seeing issue #361 and it seems that the shape/indexing problem was solved by this and is get the following hls4ml summary of the model: which seems to match up to the original model correctly.

However, now I have hit a compilation error when building/compiling the model:

I thought perhaps this related to the update going on in #402 so I tried swapping to that branch and trying again but received an identical error. Any thoughts?