Why does this tf.keras model behave differently than expected on sliced inputs?

Question

I\'m coding a Keras model which, given (mini)-batches of tensors, applies the same layer to each of their elements. Just to give a little bit of context, I\'m giving as input gr