Changed all the keywords to anonymous nodes

[jpt13653903]

@Chris44442, @superzanti,

I got annoyed with all the keywords that are explicit in the syntax tree, so I changed them all to anonymous nodes (i.e. strings, like the operators are at the moment).

But I need a second opinion... Is this a good idea? Should I merge this? Are there keywords (like all, for example, that are better left explicit?

[Chris44442]

I have tried it with a few of my vhdl files, and looked through your tests. Looks good to me, makes the tree more concise.

I don't think these keywords need to be present in the tree. Most of them are implied anyway. In other languages the tree-sitter doesn't explicitly spell them out neither (at least the ones i have tried).

[jpt13653903]

The one I'm wondering about now is the architecture declaration followed by identifier and name. I think I should clarify that with field names (I agree that removing the keywords was a good choice, but now that part of the tree is not quite as readable).

Let me come up with something and then push to this PR... I'm thinking to use the italic parts in the formal grammar for the field names...

[jpt13653903]

I'm quite happy with the commit above. It certainly makes the highlights easier:

(_ view: (_) @type)
(_ type: (_) @type)
(_ library: (_) @namespace)
(_ package: (_) @namespace)
(_ entity: (_) @namespace)
(_ component: (_) @namespace)
(_ configuration: (_) @type.parameter)
(_ architecture: (_) @type.parameter)
(_ function: (_) @function)
(_ procedure: (_) @function.method)
(_ attribute: (_) @attribute)

(_ view: (name (_)) @type)
(_ type: (name (_)) @type)
(_ entity: (name (_)) @namespace)
(_ component: (name (_)) @namespace)
(_ configuration: (name (_)) @namespace)

An architecture now builds the following tree:

(design_file
  (design_unit
    (architecture_definition
      architecture: (identifier)
      entity: (name
        (identifier))
    (architecture_head
      ...
    (concurrent_block
      ...
    (end_architecture
      architecture: (identifier)))))

I'm going to merge to develop and then we can continue testing...

[superzanti]

I think this is a good idea - anonymous nodes can still be used in most tree-sitter extensions.
I'm fixing most of my scm files now, so I'll let you know if there are any keywords that would be better lefts as non-anonymous. I'll let you know.