[PROPOSAL] Consider moving some/all of glib::wrapper to a proc macro

[jf2048]

Altogether these macros are huge, it is difficult to add features or refactor them. Every new argument must put through around 6-8 pattern matches to work, see for example some of the changes here: 3c41d6d

If we ever want to add support for lifetimes and PhantomData that will need to have all the macros changed again as per #610 (comment).

[sdroege]

@GuillaumeGomez is talking about that since quite some time, just nobody got to it yet :)

[jf2048]

Tried out here cc620d4 but it seems around 15% - 20% slower than the macro_rules implementation. Not sure what the cause of the slowness is, if it is avoidable or if it is just overhead from the proc macro bridge