Incorrect string escaping in TOML encoder

[ghost]

While encoding string values, the TOML encoder incorrecty handles backslash characters that occur in the string value.

For example:

toml.dumps({"a": "\\x"})

returns

a = "\u00"

which is not valid TOML.

This bug also triggers failures in the encoder tests "escaped-escape" and "string-escapes" from the TOML test suite.

[LexiconCode]

@uiri This bug is affecting two projects I am affiliated with Caster and Dragonfly.