Our TDML test files are XML 1.0 documents, so cannot have certain characters in them.

Those characters are sometimes needed in data, particularly if we pick up a file of textual data and try to embed it into a TDML self-contained test file. This test data may not obey XML 1.0 constraints. E.g., XML 1.0 does not allow character U+001b at all.

We could enhance the TDML runner to preprocess the <document> element of a TDML test, replacing DFDL-style character entities like %#x1b; with the corresponding character.

This would let tests be converted from separate files to self-contained TDML more easily.