First, please fix the citations, that’s not what I wrote.
Now, I still haven’t seen any convincing argument why mixing codes and human readable values as appropriate for some sign is a bad thing. You’re repeating this over and over, but I just don’t see any reason apart from shortcomings in current editors.
Sure, there are problems with long texts and non-standardized signs, but none of this is fixed in any way by your proposed new tags. The proposal currently shows only the most simple cases of single signs that are covered perfectly well by the two currently existing tagging possibilities. Complicated example where a new scheme really makes a difference are needed.