...but I'd expect it to work like a typical 32-bit x86 installation; meaning that while a single process is limited to 4GB, the OS can have multiple processes running to use the entire 8GB address space of the CPU

You are correct...

I skim read this article a few weeks back and it was only then that I realised the limitation of 4Gb RAM on a 32 bit. What I didn't properly read was:

"The 32-bit kernel is joined by a 32-bit userspace, however: While the kernel can see and use all 8GB, any given process on the system is limited to 4GB. In other words: If you’re running a single application which doesn’t fork into multiple processes, it’ll only ever be able to grow to half the 8GB available"

We 'attend' several virtual conferences and training session which are full day or multiple day events. So we take part using the big television in the sitting room. I plan to get a Raspberry Pi as streaming device for these events and it will largely just run one instance of one browser with one tab open so, for this application, I might hit the limitation. But for general usage, like the OP is looking for, it is probably the ideal solution.