Parse data types based on current dialect

Currently we parse data types ignoring the dialect, which doesn't make that much sense. This requires the caller to validate not only their usage, but if we parsed the data types correctly.

Could we add the dialect to parse the data types? That's done in a few parts of the code using `dialect_of!()`

Locate the data type parsing entry points and the existing uses of `dialect_of!()` mentioned in the issue. Trace how the current dialect reaches related parser code, then verify that data types are parsed with that dialect and that the existing parser tests cover dialect-specific behavior.

Written by the indexing model from the issue text.