I recently noticed that we have TA-IN (Tamil India) as one of the languages, also it doesnot have any teams yet. In india we use the tamil code as TA and most very less frequently as TA-IN. For srilanka tamil they use TA-LK. So it is highly unnecessary to have TA-IN branch.
You are right that ta_IN should not be used, as glibc only works with 'ta' code. It's here probably because of the convention that Publican-based documentation use. As long as there is no team created, no actual translations can be send for this invalid (for software anyway) language code, so we're fine.