* make the factory always return writers via shared_ptr
* add a PathWriter wrapper for opening a stream for the lifetime of a normal Writer
* wrap writer factory exceptions
* hide some writer factory implementation details in dteail:: sub namespace
* rename TokenWriter::create to create_stream_writer, add create_path_writer
* bump library version

Allow custom "ign" tag equivalent in tagsets, fix tests accordingly as now a valid ign tag is required

add a new ParseMode flag to the tagset class that disables exception throwing and makes tag parsing return an ign tag in case of parsing failure

tweak AnnotatedSentence wrapping functions to behave more reasonably when passed a sentence tat already is an AnnotatedSentence

add two new writers: a dummy "none" writer that disables output and a "stats" writer that outputs general info only about token counts and annotation info

Reader factory. There is some code duplication with the writer factory as well as among the individual Readers, but overall it is now easier to support more input formats.

Add Tagset::parse_symbol_string for convenience parsing of comma-separated symbol strings

Fast and drity (regex-based) XCES parser based on the one in Pantera, noticeably faster but prone to breakage on even slightly freeform xml files