here we should default to returning the original string, otherwise we risk destroying the unicode string when mbstring becomes un-available for some reason. Better keep original string than risking breaking it 👍

We aren't doing that in the method above for strtolower as well

let's do it in all mb_ proxy methods to avoid destroying unicode strings

Imho we shouldn't take care of that. As we can't avoid problems with unicode. substr for example might cut off within a multibyte char, strlen might return an incorrect length.
Also Twig for example does the same in some of their filters: if mb_strtolower isn't available they use strtolower

As we can't avoid problems with unicode.

Actually we can avoid potential problems by not touching the string... ideally we would detect when it's a unicode string but this may get tricky (or slow)

Well ok, if you think it makes sense to change that now after it was already over 6 years in the code and we didn't get any bug report about it...

I think what @sgiehl was pointing out we might avoid it the problem there but get the problem somewhere else... not quite getting the problem as well @mattab ?