Not sure you're going to find much info on the specific board in the BIOS, MR BIOS images tended to be generic per chipset.
Reverse-engineering jumpers can be quite fun, although late 486 boards like this are of course the most challenging with their sheer number of jumper. Good news here is that the jumpers seem clearly separated into separate functional groups, particularly those around the VRM. Bad news is that some groups seem to be missing jumper caps entirely (the set down by the VLB slots). Not having a known-functional setup doesn't help...
As for that asinine warning: turn it around. Consider what can actually damage a CPU and avoid that. Then you're safe to try all other combinations. Basically only overvolting can actually damage a CPU. Fortunately voltage is something you can easily measure (if you have a multimeter). Fire up the board without CPU and measure voltage between Vcc and GND pins in the socket. 3.3V: safe for that DX4 as per sticker. 5V: not safe for DX4, stick in an older 5V part.
Ideal test CPU would probably be an Am5x86-133 as they can tolerate 3V-5V for short periods and most will run at 3x50MHz so will at least boot if you accidentally set bus speed to 50MHz (assuming you didn't set the multiplier to 4x, as 200MHz working is definitely not a given). Also the Am5x86 supports WB caching, but (unlike Cx5x86) will happily run with same simple settings as i486SX/DX. Second choice would be an i486DX-50 (not DX/2), again because of 50MHz bus speed.
Apart from a known-good test CPU you also want a single, known-good FP SIMM and a known-good ISA VGA card. PCI might work. A VLB card is also useful, but given those missing jumper caps, I'd first want to confirm everything works with ISA before messing around with VLB.