cleanRomaji breaks valid double N notation

[risseraka]

Hi there,

This might be an unresolvable issue, but here goes.

Cleaning valid romaji breaks it.

E.g.:

toHiragana(cleanRomaji("DANNA"))
// DAN'A
// だんあ wrong
toHiragana("DANNA")
// DANNA
// だんな correct

I understand that the rationale behind the following code is to correct outdated writing of romaji

// Convert the NN form into the more common N'
str = str.replace(/NN/g, "N'");

but then, how was 旦那 (husband) written? DANNNA?

WDYT?

[lovell]

/cc @jeresig who originally contributed this useful feature and may have suggestions.