So, I actually rewired this with a pair of SPST toggle switches I had lying around (one for ON/OFF, one for FAST/SLOW). In retrospect, I don't think I had a switch problem in the first place. I still see no apparent AGC action, and the S-meter AGC value is near-zero all the time.
When I disconnect the Common line from the board, and measure between the Common line and ground (with everything else connected) while cycling through the switch positions, I get what is expected:
OFF - 0 ohms
FAST - ~470k
SLOW - ~1M
However, when I reconnect Common to the board, and measure the same resistances, then everything basically measures as a short through the switch. So clearly there is a path to ground somewhere, regardless of switch position, and I figured I would check out the diodes. I measured both forward and reverse across both diodes; in both cases I get about 550 mV forward and 1100 mV reverse. At first I thought maybe this was an issue with the diodes, but looking at the schematic, it seems like there is probably always a return path via Q3 gate-source and the other diode. (Looks like the 2N7002's have about ~690-720 mV drop from gate-source?)
So I'm thinking a DC resistance measurement across the switch positions in-circuit doesn't really work due to the presence of Q3? (I.e. I'm really just measuring some gate-source resistance)
Any other thoughts?
So, still stumped. My switches are wired up right.