Improve type parsing to support arrays and generics

[metoule]

Note that I had to remove the token QuestionDot and QuestionOpenBracket, because it prevented the proper parsing of int?[].

It's now possible to write:

var target = new Interpreter();
target.Reference(typeof(IEnumerable<>), "IEnumerable");
target.Reference(typeof(Dictionary<,>), "Dictionary");

Assert.AreEqual(typeof(IEnumerable<int>), target.Eval("typeof(IEnumerable<int>)"));
Assert.AreEqual(typeof(IEnumerable<IEnumerable<int?[]>>), target.Eval("typeof(IEnumerable<IEnumerable<int?[]>>)"));
Assert.AreEqual(typeof(IEnumerable<>), target.Eval("typeof(IEnumerable<>)"));

Assert.AreEqual(typeof(Dictionary<int, string>[,]), target.Eval("typeof(Dictionary<int,string>[,])"));
Assert.AreEqual(typeof(Dictionary<int, IEnumerable<int[]>>), target.Eval("typeof(Dictionary<int, IEnumerable<int[]>>)"));
Assert.AreEqual(typeof(Dictionary<,>), target.Eval("typeof(Dictionary<,>)"));

Assert.AreEqual(typeof(int[]), target.Eval("typeof(int[])"));
Assert.AreEqual(typeof(int?[]), target.Eval("typeof(int?[])"));
Assert.AreEqual(typeof(int?[,,][][,]), target.Eval("typeof(int?[,,][][,])"));

Fixes #172

[davideicardi]

Thank you @metoule!

Are you able to fix the conflicts? (I have merged the other PR...)

[metoule]

@davideicardi I've fixed the conflicts :)

[Pentiva]

As far as I can tell, this implementation does not support types with a variable amount of generic type parameters?, e.g.

var target = new Interpreter();
target.Reference(typeof(Tuple<>), "Tuple");
target.Reference(typeof(Tuple<,>), "Tuple");
target.Reference(typeof(Tuple<,,>), "Tuple");

This can be worked-around by naming them Tuple1, Tuple2, etc.
This might need too much to be refactored to be done in this pull request?

[metoule]

@Pentiva that's a different issue: the goal of this PR is simply to support parsing generic types in a typeof expression.

Your issue is that it's not possible to reference multiple arity of the same generic type name. That's because under the hood there's a dictionary that maps a type name to the actual type, and therefore the last call to Reference overrides the previous one. That needs to be modified, but it seems out of scope of this PR.

Do you mind creating a new issue?

[Pentiva]

@metoule I expected as much, just thought I would mention it in this pull request as it is sub-feature of generics and, until this is merged, falls under the adding generics issue. I'll create a new issue.

[davideicardi]

Great!