Summary: The resource per process issue is isolated to just the transaction isolation mechanism and can likely be improved but you are never getting around the context switching and memory locality issues of the per process model.
Not really. The page-table overhead of a 100k processes alone is going to a eat up a lot of memory - and you don't have that with threads. And with something like postgres the page tables aren't going to be tiny, even with explicit use of huge pages for applicable allocations. The decrease in TLB hit ratio also is very painful. In a lot of workloads postgres spends way too much time in TLB faults, and that's to a good degree caused by the process model (also caused by linux making it way too hard to map executables with huge pages, itlb faults are a significant issue).
That's not to say that a 1 connection: 1 thread approach is the right thing for a pooler - threads area also don't scale great. I think there are good simplicity arguments to be made for a 1 connection: 1 thread model for a database server itself, but for poolers the benefits seem much smaller.
Your analysis seems pretty on-point with what I expected, i.e snapshot scalability and other postgres specific arch being the major bottlenecks for higher connection counts.
Instead the poor scalability stems from the amount of resources allocated to each connection. If postgres was to switch to threads without changing this architecture it's likely scalability wouldn't be majorly improved.
By the same token a connection pooler that is fork based but only allocates a very small amount of resources per connection is still likely to a net win as it can do statement based pooling of a smaller number of postgres connections.