Fix issues with unicode ANSI skipping

[ztravis]

The current implementation of unicode ANSI replacement sequences is slightly wrong - according to https://www.biblioscape.com/rtf15_spec.htm

• A scope delimiter (i.e. "{" or "}") should end the current skippable data
• Control words or symbols should be considered a single skipped character (and in my testing with MS office, they're ignored)
• Any binary data is also considered a single skipped character