This third release makes the library ready for experimentation, but not for production use yet.
A lot has been implemented:
* auto-generated IDs, enabled by default
* support for reading from stdin and outputting to stdout instead of reading from/to file
* searching in working directory when loading/saving files
* add multiple annotations from a json file (annotate_from_file())
* merge multiple annotation stores into one
* powerful regular expression based text search via search_text()
* unicode character position to utf8 byte conversion
* configurability via the Config class
* Serialisation/deserialisation to/from stand-off files via the `include` mechanism
* Renamed to_string() to to_json()
* Ordered iteration over text selections in a resource
* a position index that aids in mapping text offsets to annotations and enforces an ordering
* Initial support for internal ranged selectors, improving the space efficiency of complex selectors (annotation/stam11)
* More efficient unicode points to utf8 byte conversion
* Improved JSON error feedback, and error feedback in general
* Many bugfixes and major refactoring