Thank you for the explanation but I think I still do not get the point.
Your suggestion to extending the knowledge about the datatypes sounds interesting and I think I have seen some limitations regarding unicode or not and size of datatypes in the past. So I would like to see enhancements in this area but I guess it is a kind of a bigger job to implement these things.

Do I understand it right that the unsupported use case looks like this:
1) oracle database between Version 8 and 12, so all supported versions are affected
2) database table with two "binary" array columns configured via hbm file
3) one of the built in oracle dialects
In this case hibernate cannot create a valid schema for the database on oracle using the built in dialects because it trys to use a datatype that is not to be used twice on oracle. And it does not detect this problem and does not display a warning. All other databases do not have this problem.
I mean I understand the backwards compatibility argument but if I understand it right this case is not supported and it should be, right?

I can easily go on using a derived dialect but hope I make myself clear this has a kind of bad smell to use this approach from my point of view. I did not find documentation about using your second suggestion overriding BasicTypeRegistry. I am not familar with that.
However as soon as I find some time I try to dig a little bit deeper using the hints you gave to me. So thank you for that..