Fixing the CI for aidge_quantization

This merge request aims to fix an error in the functional testsuite of aidge_quantization [#85 (moved)\]

Specifically, the recent addition of a second output to the MaxPooling2D node has caused the convnet model to produce random outputs.
To address this, I modified the propagate function in the test script to exclude any outputs that may originate from a MaxPooling2D node.