all: make type of mut function args and loop values consistent

[UweKrueger]

Function arguments, receivers and for loop variables that are declared mut are references behind the scenes. However it is not clear what their logical type is, i.e. if the behave like values or like references. The present state is like this:

fn f(mut y St) &St {
    println(typeof(y).name) // -> &St
    mut x := y // `y` behaves like a value, i.e. `x` is of type `St`
    mut z := [y] // again `y` behaves like a value, i.e. `z` is of type []St
    return y // here `y` behaves like a reference (`return &y` would not be allowed)
}

This is an inconsistent behaviour, which is made possible by a bunch of exceptions in checker and cgen.

This PR changes the state of mut arguments so that they are logically pure values. In particular typeof(y).name would become St and return &y would be required. This eventually reduces the number of exceptions required in checker and limits them to cgen. (At the present state there are some exception that had to be introduced in checker to make existing code continue to work, but these will become warnings and can be removed later.)

This fixes #11038.

[SleepyRoy]

@yuyi98

[yuyi98]

@UweKrueger I hope the conflict can be resolved as soon as possible and the merger can take place as soon as possible!

[JalonSolov]

Only CI failure was in MacOS compiling V UI - that's been fixed now, hasn't it?

Taking care of the conflicts should allow this to be ready to merge.

[UweKrueger]

Taking care of the conflicts should allow this to be ready to merge.

@JalonSolov Unfortunately it's not that trivial. A lot of code has been contributed in the last few weeks that causes semantic conflicts.
In particular there has been significant progress in the JS-backend (thanks to @playXE 👍) that was done with the old behaviour in mind.
For this reason I'm reconsidering my approach: Even though the checker does its checks with the logical types it should be somehow possible to put the low-level type into the AST. That would minimize the changes in the backends...

[danieldaeschle]

Any progress? :/