run forward pass even for empty subset to produce correct layer outputs

[avtc]

@Qubitium
This is a simplified fix for #2116, maybe it is not optimal, but it works.

The suggested solution is too hard for me to make it properly:

If a full layer has no module to quantize a simple forward() is enough and output is captured to be used as next layer's input. So one pass forward (entire layer simple forward wihout need of dealing with subset loops and micro forward loops, just full layer, usally XXXDecodeLayer.forward(). So output = current_layer.forward() is enough or sometimes just calling the layer callable like layer() which same as layer.forward().

Assume layer 2 has no modules to quantize. At beginniing loop for layer 2, we have layer_output from completed forward_replay() of layer 1. Then pass this to layer 2 (as a whole) as layer_input and store ouput, then immediately loop to layer 3 without any further subset work that is only necessary if we need to quantize part of a layer.

And it looks it is correct for the case when one or more subsets of a layer are excluded from quantization.