The error was saying that "infinitely many alphabets can be generated" but there is a finite number of parameters which have a finite range of values. If Knuth had just written "a humongous number of alphabets" he would be correct.
Sure. For practical purposes, the number is infinite. This is why I found the bug and the bug report so amusing: In "normal" language, the sentence is correct, but for a mathematician, especially such a renowned professor, it is an amazing blunder. Bonus points for being the first word in the document.
I would say he could have even gotten away with “functionally infinite number of alphabets”.