…ode points

The internal string representation is changed from UTF-8 with replacement
characters to a modified form of "WTF-8" that is able to distinctly encode
UTF-8 errors and UTF-16 errors.

This handles UTF-8 errors in raw string inputs and handles UTF-8 and UTF-16
errors in JSON input. UTF-16 errors (using "\uXXXX") and UTF-8 errors (using
the original raw bytes) are maintained when emitting JSON. When emitting raw
strings, UTF-8 errors are maintained and UTF-16 errors are converted into
replacement characters.

…rings

UTF-8 errors and UTF-16 errors that were previously encoded into the ends of
strings will now potentially be used to form correct code points.

This is mostly a matter of making string equality behave expectedly, since
without this normalisation, it is possible to produce `jv` strings that are
converted to UTF-8 or UTF-16 the same way but are not equal due well-formed
code units that may or may not be encoded as errors.

Errors are emitted as negative code points instead of being transformed into
replacement characters. `implode` is also updated accordingly so the original
string can be reconstructed without data loss.

This is no longer needed as strings are capable of storing partial UTF-8
sequences.