Implicit token highlighting

[fan-tom]

Motivation

Currently both implicit and explicit tokens (terminology from here) share a common highlighting key: BNF_TOKEN, which makes it impossible to assign different highlighting styles to them and spot issues in the grammar, when e.g. because of a typo implicit token is used instead of the name of an explicit one.
E.g. this grammar

{
  tokens = [
    OPEN_PARENTHESE = '('
    CLOSE_PARENTHESE = ')'
    ID = "regexp:\w+"
  ]
}

file ::= OPEN_PARENTHES ID+ CLOSE_PARENTHESE

Is highlighted like this

This PR makes it to be highlighted like this (by default, may be overridden on the color settings page)

which lets developer notice that something's off, and overall distinguish explicit tokens, matched by name and implicitly defined tokens, the same way we already distinguish explicit tokens, matched by value and toekns, matched by text (called "patterns" in the code base)

Additional changes

There are 2 sets of commits:

• those which introduce separate a highlighting for implicit tokens, with minumum changes across the codebase
• those I considered to be nice to have (they have fix: prefix):
  • let IDEA highlight bnf example text on the color settings page as an HTML
  • use more of instanceof pattern matching syntax where I already started using it
  • rename TOKEN to EXPLICIT_TOKEN to make it symmetric with IMPLICIT_TOKEN

Please advice which of them are desired and which I must drop

[YannCebron]

Thanks a lot for this!

[YannCebron]

changing the identifier will break any customizations users already have, please keep "BNF_TOKEN" as ID

[fan-tom]

fixed