Expose `ts_parser_set_encoding` on the Parser API

[saket]

Tree-sitter's C API supports ts_parser_set_encoding, which makes node offsets line up with utf16 code units instead of utf8 bytes. This is a natural fit for JVM languages, where String is already utf16 internally.

ktreesitter v0.24.1 hardcodes utf8 in both parse paths:

parse(source: String)
parse(oldTree, callback)

This forces Kotlin callers have to maintain a byte to char offset table on every parse to bridge tree-sitter's utf8 offsets to Kotlin's utf16 string indices. Exposing a knob would let us skip that entirely.

Is there interest in this? Happy to contribute a PR if the direction sounds reasonable.

[ObserverOfTime]

I think it would be reasonable to use UTF-16 in JVM & Android and keep UTF-8 in the Native implementation.