Last time, the calculated beam waist I got was based on two identical mirrors with no reflecting coating layers on the back mirror. When that layers are taken into account, I got the same waist size.
The new plot is attached below. Two sidebands are 21.5 and 35.5 MHz. The peaks, from left to right, are 9th, 6th, 3rd, 28th, 25th, 22nd, and 19th order. Choosing 35.5 MHz for frequency modulation should be fine.