I agree very much, that yuor '51 MUST be bypassed, both with a 100 nF ceramic and with 4.7 to 10 µF electrolytic cap, but why tantalums?
Aluminum, I am sure is fine, I just do not have the room for them and thus have not tested.

"Put the thing in a steel box"
Okay, but should'nt it also be a magnetic box?
That is why it say "steel", not "metal"

"Do not drive capacitive loads (like ribbon cables) directly from Uc outputs, buffer them with TTL or CMOS buffers (140-144)"
What about using a pull up resistor. The datasheet for 80515 states that "high" is 2.4V with 80 µA behind it. Thats a verry high Z input. With a pull up in order of 5-10 K, (at Vcc 5v) you should be far better of.
Yes, but a buffer can drive 50 mA and hold it against noise spikes, a resistor will allow noise to come through, just round and shorten the pulse. Also, adding a passive pullup will reduce the chips ability to "hold" a low.

Ha' en rar dag

Erik