[Bug c++/86943] [7/8/9 Regression] Wrong code when converting stateless generic lambda to function pointer

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86943

--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> ---
I might be wrong about the C++14 rules ... maybe the {} initializer should
directly initialize the parameter, without any move.

I don't know if we need to fix the incorrect argument to the move constructor
as well as prevent that move, so the incorrect move doesn't remain latent.