We're actually doing this right now
In our project we're collecting 3DGA data from children with CP walking barefoot and with two different types of custom-made hinged thermoplastic ankle-foot orthoses. We've drilled holes in the ankle-foot orthoses allowing the EMG electrodes to protrude through the orthosis with a margin of clearance around each electrode to avoid movement artefacts. This method was deemed most suitable because we have an older myon wireless EMG system where the transmitter "heads" build quite a bit. I think the newer models build significantly less but may nonetheless incur signal disturbance if subjected to relative motion from surface electrode to orthotic device. I know that other EMG+AFO studies have collected data with electrodes placed under orthotic devices so this method has definitely been employed before but we weren't able to get good signal quality with our equipment.
Would love to hear more about your project!

Best regards,