Functions Transliterate, Fold, Clean, the ranges they cover in order?

[seaspeak]

I've summarized from the website explanations:

fold outputs are all ASCII characters. (<=)

you may wind up with non-ASCII characters in your citation keys. You can prevent that using the fold function at the appropriate place in your pattern, but checking this checkbox will just apply fold to all your keys.

Then it implies clean do more than transliterate. (>=)

clean transliterates the citation key and removes unsafe characters.

the example extra('tex.shortauthor').transliterate.clean.lower.len + year, it cleans after transliteration.

But here it seems to say clean equals transliterate

Note: The functions above used to have the clean function automatically applied to them, this is no longer the case, so if you have CJK authors/titels and you want to manipulate them (using eg. capitalize), you could have to use transliterate on them first, eg. authEtal2.transliterate.capitalize + year + shorttitle(3, 3).

It suggests to use fold anyway.

you could use something like Auth.nopunct.skipwords.fold after adding jr to the skipWords list.

so the conclusion is fold > clean > transliterate? Acutally, I don't notice any difference.
If I untick the option "Force citation key to plain text" what function is best for Latin alphabets. Should I use fold, clean or transliterate?