Merge: Named arguments

Named arguments allows the caller to specify some subset of arguments it wants to call.
This PR introduce a syntax for callers inspired from the Python one (the Ruby one uses columns that clashes with types declaration).

~~~nit
format(linewidth=80, separator=":", indent=true)
~~~

The specification tries to be as simple as possible

* Only default parameters can be passed by name; because the other are mandatory.
  This restriction is also useful to avoid to specify additional rules, like the last parameter of assignment methods, because the ones on default arguments apply.

~~~nit
fun foo(a,b: nullable Int) do ...
foo(b=1)
foo(null,1) # equivalent
~~~

* Named arguments can appear anywhere (not necessarily at the end)

~~~nit
fun foo(a: Int, b: nulable Int)
foo(b=1,2)
foo(2,1) # equivalent
~~~

* The order of evaluation is the order of the arguments (not the order of the parameters).
  I think this is more POLA this way.

* The names to use are the one of the statically designated method declaration (not the introduction).
  This one is problematic because the language permit the name of parameters to change in redefinitions.
  But in case of errors or in a IDE it is often more useful to point the most precise method definition.

The implementation was trivial because the hard work was already done in the previous PR.

This PR was unlocked by #1280

Pull-Request: #1281
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Reviewed-by: Etienne M. Gagnon <egagnon@j-meg.com>
Reviewed-by: Lucas Bajolet <r4pass@hotmail.com>
Reviewed-by: Romain Chanoir <chanoir.romain@courrier.uqam.ca>