Related a bit to https://mantis.opcfoundation.org/view.php?id=10342.

When the StructureDefinition is used as the DataTypeDefinition Attribute in normal client-server address space, the Abstract-ness information is available from the IsAbstract Attribute. However, when in https://reference.opcfoundation.org/Core/Part14/v105/docs/6.2.3.2.2 DataTypeSchemaHeader, the IsAbstract Attribute isn't available, since there is no address space. That is used as a base for UaBinaryFiles and PubSub metadata.

In the StructureDefinition https://reference.opcfoundation.org/Core/Part3/v105/docs/8.48 there is the DefaultEncodingId NodeId field. It now says:
"The NodeId of the default DataTypeEncoding for the DataType. The default shall always be Default Binary encoding.
If the DataType is only used inside nested Structures and is not directly contained in an ExtensionObject, the encoding NodeId is null.
The DefaultEncodingId in the StructureDefinition shall always be the Default Binary encoding for the DataType "

I think if it could be added that "If the DataType is abstract, the DefaultEncodingId field shall be null", it would be possible to depend on this to know abstractness of a type.

Note that this is a "futureproofing" type of an issue. Currently there isn't a direct need. But there might once the reader of a UaBinaryFile or a Subscriber in PubSub would need to instantiate something based on the types it might be needed.

I should note that DataTypeSchemaHeader uses StructureDescription, which wrap the StructureDefinition (+DataTypeId + Name). I do not see it being changed at this point, unfortunately this was missed when it was made (but would have been the logical place to add this information).