TEL parser tweaks + standardized AST nodes (to abstract ANTLR from users)

[dvdotsenko]

What Changed:

• split g4 into separate Lexer and Parser files (originally to allow reuse of Lexer file in more ambitious Parser. That parser not included in this PR).
• converted majority of expr into left, operator , right format to simplify Visitor logic.
• Embedded a markup-to-AST Visitor into this repo (to allow simple reuse in non-Diesel code).
• Expanded tests from "does not blow up" into "let's see that it parsed the right thing
• added support for IN , LIKE ILIKE to TEL expressions. (these are allowed as Filter expressions in Blending Husky and are very commonly used by users in normal TEL).
• added syntax error detailing logic that highlights the issue in fragment:
  pql_grammar.from_pql.ParseError: Unexpected "ilike" (positions 3 to 7) in fragment "a -->ilike<-- 'asdf'" (See TelErrorListener - allows init without need to pass input stream to it.

Goals:

• add "standard" markup-to-ast visitor to TEL_Grammar repo to allow simple reuse in scripts.
• "standardize" simple / serializable "light" AST for expressing (NOT validating, NOT transformation) the contents of TEL, basic traversal through it, tools to find nodes in it easily - all to help in reuse of this code in stand-alone scripts outside of Husky.
• add support for extra keywords while we are at it.