All that does is complicate the already complicated EDL phase by an order of magnitude.
Trying to launch a drone from the lander during descent would mean:
1) the drone is able to begin its power up process at the end of the cruise phase while the whole probe is buttoned up for reentry
2) the drone has a big enough battery to power avionics, comms, component heaters, and the motors
3) the drone carries a big enough radio and antenna to at least reliably communicate with one of the orbiters relaying data from it and the main spacecraft
4) there's a at least one orbiter positioned to receive data from the drone and the drone manages to perfectly aim its highly directional high gain antenna at the orbiter after an unknown launch position and orientation
5) everything goes perfectly with the drone and it doesn't collide with or damage the sky crane or rover or debris from its launch (or failure) doesn't damage anything
Saying it "increases risks" with the landing is a vast understatement.