Making sure I have the appropriate hardware for a SSerial-based machine

If the firmware can generate an SPI bus on the expansion header, there is no need for any of that...the input isolation chip comes in an SPI-output version that can be chained for up to 64 isolated inputs on one SPI bus.

On top of that, the SPI version of the chip has programmable debounce filters and high-speed digital capable inputs, but the parallel output version I used isn't programmable, but can be had in various debounce filters and high-speed/normal-speed inputs.

Once I get one of these working and verified, I'll try to develop a more capable version, the plan is to eventually get one going that will speak SmartSerial to drive an automatic tool changer.

SPI is possible but a real mess because the hal driver requires chip specific
code. Its also not the best for longer signal runs (cables) unless both ends use a slow data rate and filtering.

How about an intermediary, like an ATtiny or something as a go-between for SPI to whatever?

I'm using the SI8380P in this first board, the parallel, no-debounce, bare-bones version. The SI8380S is the 8-input, SPI-capable version.

That might make sense if the host interface is sserial
(which hides the hardware implementation)

I figured that might be the case. In that situation, an STM32 using the STMBL-based code is probably the most straightforward route.