feat(postgres): parse CREATE TYPE statements

[devfros]

• Solves #7612
• Adds AST parsing for Postgres enum and composite CREATE TYPE
• Reuses exp.Create, exp.DataType, exp.Schema, and exp.ColumnDef
• Adds Postgres dialect tests

[VaggelisD]

Hey @devfros, nice work! Leaving a couple comments from my initial pass:

[georgesittas]

This is looking better. Left a few more comments to clean it up even more. There may be an opportunity to support the whole syntax spec in one fell swoop.

[georgesittas]

Do we actually need this fallback? It seems that the non-AS syntax is:

CREATE TYPE name (
    INPUT = input_function,
    OUTPUT = output_function
    [ , RECEIVE = receive_function ]
    [ , SEND = send_function ]
    [ , TYPMOD_IN = type_modifier_input_function ]
    [ , TYPMOD_OUT = type_modifier_output_function ]
    [ , ANALYZE = analyze_function ]
    [ , SUBSCRIPT = subscript_function ]
    [ , INTERNALLENGTH = { internallength | VARIABLE } ]
    [ , PASSEDBYVALUE ]
    [ , ALIGNMENT = alignment ]
    [ , STORAGE = storage ]
    [ , LIKE = like_type ]
    [ , CATEGORY = category ]
    [ , PREFERRED = preferred ]
    [ , DEFAULT = default ]
    [ , ELEMENT = element ]
    [ , DELIMITER = delimiter ]
    [ , COLLATABLE = collatable ]
)

CREATE TYPE name

I think we may be able to "just" parse these with the existing logic as "properties", without having to add any more logic.

[devfros]

Removing that fallback regressed tests, unsupported CREATE TYPE forms no longer roundtrip as Command. The idea was to to scope this PR to AS ENUM / AS (...) and avoid partial ASTs