The concern I had was, based on the length of the delay being 20 seconds, the values of R & C would become rather large. I thought of using a simple 555 timer but it has similar issues with long time delays and the timer wasn't really designed to be a one shot.
GM went the route of using a 14 stage 4060 ripple counter to allow for a faster clock pulse but then had the issue of having to stop the counter once the final count was reached. This added morel circuitry. They were seriously limited to components available in the late 80's. They also had cost constraints factored in as well, so I'm sure they used the best approach they had available.
Today we have a much larger selection of components and costs are much lower as well. Especially at higher volumes. I build circuit boards in batches of 3 from a prototype vendor (Osh Park). The board for this project came in at $21.50 for a batch of 3 or approximately $7.00 each. The micro controller was around $1.20 each. Actually cheaper then the various capacitors, resistors and transistors that GM used in their original design. This allowed me to "splurge" and build several different designs and test them out. The final design I ended up with is basically a ATTiny85 CPU, a voltage regulator, a single transistor driver and a relay to make the connection. All the other parts were for voltage spike suppression, and visual verification of relay function. The headers at the top left are there for programming the CPU while in circuit on boards I don't put a socket on. Since these are still prototypes I'm socketing the CPU to be able to program it by itself. Final design will have the CPU soldered directly to the board and probably wont have the programmer header pins installed since it's doubtful that any end user will need to modify the program.