Wow! That's a cut down board. It's crazy enough that you should submit that picture to theretroweb.

The marks on the back near the CPU socket look like they could be electrical burns. If that's the case, there might be some deeper issues than the marks on the surface.

The markings on the VLB slots are curious. Master / Master / Slave

Does that indicate which slots can take bus master controllers or which should be populated first?

I managed to "fix" the board.

There are still some issues which appear from time to time... probably some fatigued soldering joints, as I'm unable to pin-point them.
This issues appear way more less when the "T(ur)B(o) SW"/JP36 is closed. Tested with a 486DX/33MHz and a 486DX2/66MHz
Issues:
1) Boot process halts at d3 bios code. Re-seating the memory modules may or may not fix the issue, but doing this few times in a row usually allow the boot to continue. I resoldered the memory sockets but still having this issue.
2) Keyboard is not recognized. Rebooting again seems to fix the issue until some later time.
3) Video is monochrome. In the same session could be monochrome in 3DBench but color in PC Player Benchmark for example... no clear pattern here.

I repaired some traces on the board because they were broken.
The traces around the processor were repaired with some thick wires (I don't have a microscope )
The 82C602A chipset had a broken trace under a power pin and I replaced it with a wire.
I resoldered the 82C895 chipset as multiples pins didn't make contact with the PCB.
I replaced all soldered config jumpers (except those for cache config) with pin bars.

You can try checking all resistor ladders, maybe some are on address bus and cracked one would match symptoms.

I checked all resistor ladders and they are all good and in specs.
Also i resoldered 82C602A chipset and no improvement observed.
Two new behaviours appeared:
Sometimes a Keyboard error is displayed on boot but the keyboard works between 1 and 5 keypresses and then is unresponsible.
Sometimes the boards restarts for 2,3 times before entering the runnning state.
What i suspsect now is that one of TTL logic chip is faulty.

Those marks suggest board was stored under a big heavy pile so mechanical damage is very likely. I would just start refreshing solder joints on everything.

have you tried applying slight pressure on the pcb while trying to boot to see if there is a spot it likes/doesnt like?