[Hermes Grammar] Implement string escapes for #247

[cjllanwarne]

Grammar changes required for #247 (except the part about WDL being a UTF-8 document) EDIT: the parser seems to accept UTF-8 encoded documents - hooray!

Demonstrated in action in broadinstitute/cromwell#4427

• Captures escape sequences properly (previously we were missing \")
• Makes escape sequences a token type rather than silently converting them.
  • This should make highlighting and conversions easier.
  • It also meant I didn't need to mess around in the language-specific wdl_unescape functions to allow the (previously unimplemented?) \xAB style and U000000AB style escape sequences.
  • But engines will now need to process escape tokens to make the final strings.

NB: a similar change would be required in the WDL 1.0 parser if we want to make the above fixes (since the same bugs are also present in the 1.0 grammar)