Type inference: map + accumulate tuple recursively

[tkoolen]

I’m guessing because in the second version it’s easier for the compiler to prove that the recursion is finite.

This reminded me of Jameson’s post here: Efficient tuple concatenation - #8 by jameson and explanation here: Efficient tuple concatenation - #11 by jameson

@jameson

It doesn’t call itself recursively on new values, only existing ones. This allows inference to trivially prove that it won’t need to solve the halting problem in order to do constant propagation.