GitLab

- setting config of the tokenizer is now done as an array, supporting the following attributes:
  -  maxDepth (int)
  - disableAutoTrim (bool)

- input-blocks can now be configured through the new bool $splitAffixIntoSymbols
  paramter to generate prepended and appended Tokens instead of prefix/suffix
  attributes, allowing easier parsing of naturla-language like syntaxes

- fixes bug, where not-tokenized blocks could contain other blocks

- use ResultSymbol::asGuessedType() to get the symbol-value as a
  dynamically casted datatype

- blocks which had shouldTokenizeContent disabled were returning flat ResultSymbols
  the new behavior is to return a ResultBlock of the coresponding block-type,
  but keep the block-symbol untokenized, so the block always only contains a
  single string symbol

- casting the tokenizers result to an string automagically allows
  rebuilding a syntactically correct representation of the
  original input string

- extracts tokenizer from ricwein/templater project