readJSON should always return valid, safe UTF-8 strings

[rajsite]

This is important for the Flatten and Unflatten JSON functions as they talk to external services

This is also important for interoperability with other C APIs expecting to receive valid UTF-8 byte sequences when using the readJSON function.

related discussion: #283 (comment)

It looks like when reading strings from LabVIEW, replacement characters are used on desktop. However the block diagram Flatten to JSON function instead appears to use code points U+0080-U+00FF with the same value as the byte as discussed as option 3: #283 (comment)

With C3BF corresponding to codepoint U+00FF

In order to closely align with desktop that means FlattenToJSON and UnflattenFromJSON should replace with a code point that has the same value as the byte. Edit: Should check if LabVIEW NXG behavior is the same, a quick test showed differences not fully characterized.