This is really easy.
Look at the bottom of your cpu.
Three sets of pin headers, left center, and right.
The left-hand headers are for your cabinet "direct" switches (the coin door buttons, and coin switches) and the other "in cabinet" switches, namely the start button, tilt bob, and slam tilt.
The center headers are wired the same ... all are the column drives.
The right headers are also all wired the same ... all are the row returns.
The way the switch matrix works is, the game sets each column drive, one after the other, column 1 through 8, from "high" to "low" ... or +12 to ground.
The row returns (and indeed, any other switch input including flipper buttons and direct switches) are simply sitting there looking for the input to be connected to ground.
The "matrix" is simply an arrangement of switches, with diodes, such that each column "activates" a set of 8 switches, and any closed switch, the ground signal from the column goes through the switch (and its diode), back into the row return input. The game then reads the signal, and knows the swtitch state for that column's switches.
Then it activates the next column, then the next column, lather rinse/repeat in an endless loop.
So ... for this test, you want to use the coin door buttons to put the game into "switch edges" test mode.
Then unplug all the plugs from the bottom of the cpu -- this ensures nothing is connected to any switch inputs.
Now touch a jumper wire to pin 1 on the center set of headers, the switch columns.
Touch the other end to pin 1 on the right-side set of headers, the row inputs.
You should see the appropriate switch (11) show on the display.
Now move to pin 2 of the right-side set of headers, all the way to the last pin.
Then do the same procedure, but go from pin 1 on the right side headers, to each pin on the center headers.
In this manner, you are testing each and every possibility of row/column.
This would eliminate the CPU as the issue, if it passes.