Based on #1498

The main idea would be to add UUID as natively supported with fallback to binary(16) or alike where needed.
So all you need to do is to specify uuid as native type and it should use the correct/best way as per DB type.

Does that make sense?
Would this be the way forward instead of defining it more manually?
The more manual way would probably have to be to support char(36), binary(16) and alike as low level definitions.

If we go with uuid and native support as per DB type:

• Postgres supports it as native type (129bit) directly since v8 afaik
• Mysql can use binary(16) under the hood, just as other ORMs do - e.g. here
• ?

And feedback/ideas welcome.