Decoding UTF-16 Strings fails sometimes

[cameronleger]

I've found at least one recent Preset where the current method of reading a UTF-16 String in the PREPEND_LEN mode will fail. I have a fix and could PR, but I'm not sure I fully understand the situation so I'm starting with an issue.

In this case, the UTF-16 String will only decode in Big-Endian mode, and for some reason the default is Little-Endian. I don't know how the defaults or how Bitwig saves them are chosen.

My quick solution here is in read_str and to adjust the decoding portion to:

					str_bytes = self.read(str_len * (char_len))
					try:
						return str_bytes.decode(char_enc)
					except UnicodeDecodeError:
						return str_bytes.decode('utf-16-be')

It's not the best fail-safe, but in this case it's working for the existing and problem Presets.