take parameter attributes into account in CreateOptionalRefArg

[TymurGubayev]

Problem

#1048: Interop method invocations with OptionalAttribute aren't converted

Solution

• Any comments on the approach taken, its consistency with surrounding code, etc.
  I decided to leave compiler errors in code, so that BC30455: Argument not specified for parameter... becomes CS7036: There is no argument given that corresponds to the required formal parameter....
• Which part of this PR is most in need of attention/improvement?
  1. I couldn't quite achieve above goal: VB.NET doesn't respect the Optional parameter (because it doesn't need to, I guess), so we have 3 compiler errors in the VB.NET code instead of 1 should we reference a C# assembly. This means the output generated looses some compiler errors -- I'm not quite sure if that's ok. We'd need to setup two different projects to have a proper test, which I'm not too keen on doing because of the problems with Handle C# ref return when converting VB->C# #1116.
  2. For parameters without DefaultParameterValueAttribute, I use default as initializer. I think this is ok...
  3. I'm using _semanticModel.Compilation.GetTypeByMetadataName("System.Runtime.InteropServices.DefaultParameterValueAttribute"); to get the attribute class (to keep it consistent with other places). If it fails to find one, I'm assuming it's not present -- which isn't necessarily true, I think, if at all possible. Alternative would be something like this:

   //var attributeData = p.GetAttributes().FirstOrDefault(a => SymbolEqualityComparer.IncludeNullability.Equals(a.AttributeClass, defaultParameterValueAttribute));
   var attributeData = p.GetAttributes().FirstOrDefault(a => a.AttributeClass.ToDisplayString(SymbolDisplayFormat.FullyQualifiedFormat) == "global::System.Runtime.InteropServices.DefaultParameterValueAttribute");

  This is IMO marginally safer, removes a call to GetTypeByMetadataName, but is maybe uglier.
• At least one test covering the code changed

[GrahamTheCoder]

Yep looks good. At some point it'd be good to PR an udpate to the changelog, but obviously while there are several PRs flying it will just cause unncessary conflicts so it's fine for now