The test indicator measures a swept arc. They are prone to what machinists oddly call cosine error. I do not agree that it is an error at all, it is a fact of life and is easily resolved in vector components. So it is true that a DTI can't be calibrated for linearity, but there are plenty of ways to write a cal routine for a variety of uses based on knowledge of the instrument.
Most of us use it as an indicator, to indicate and locate a difference, not (necessarily) to quantify one. That's why it is a setup tool, not a measuring tool in principle.
In a way, metrology is the science of mincing words.