Yes, I know, but my point is that whoever wrote the algorithm didn't understand how hangeul works. It's standard to code in a constraint that makes sure that you spit out an actual syllable, because there's a limited number of possible syllables, and I'm surprised that there exists a transliteration algorithm that doesn't incorporate this basic rule. The fact that Google Translate didn't process ㅌ is a truism because you never actually write just ㅌ in hangeul. It's a letter; in hangeul, letters are always arranged in syllables.

The Copyright Association is using Revised Romanization, which is the official romanization system and is the most representative of actual orthography. For the most part, you can go straight from Revised Romanization to hangeul if you know how the mapping works. The Copyright Association website does look like it has the occasional mistake--"th" doesn't exist in any romanization system--but it's definitely not using McCune-Reischauer or Yale.