The output will be some kind of op amp amplifier/low-pass filter? So, figure that at 10 kohm, and switch impedance matching at 1/10 of the switch impedance: you'll get to parts-per-million with 1 ohm switches. If your switches are paired/matched, maybe 10 ohm. If you can tolerate a megohm or more (like, noninverting op amp gain stages) then 1k ohm. Interwinding capacitance can be an issue, but your frequency range is well below the 1GHz 'standard-grade' transformer-based mixers.
They have symmetry that resistor pairs can only dream of, and DC errors that shame op amps. Mini-circuits mixers have 'em already wired and put in a shielded box, for reasonable prices.
A '4046 and the dual flip-flop can easily phase-lock to follow your master signal, and doesn't cost as much as the extra signal cable.
So far, the switches recommended have included CMOS, NMOS, HEMT, transistors (in the Gilbert cell aka Howard Jones), and diodes (in Mini-Circuits mixers). They'll all work fine at these frequencies. I'd recommend against PMOS because it has higher channel capacitance, and against HEMT because it's overkill.
The switch timing is maybe off a few nanoseconds, or drifts a nanosecond: for a sinewave signal component, the error vanishes when you switch at the zero crossings, so it's a second order effect at most.