f starting (row/line, column).

The token position is used only for error reporting and can be None or
empty.

Raise ParseError on errors. The ParseError.args is a tuple of:
(token_string, position, error message)

You can use this tokenizer as a base to create specialized tokenizers
for your custom algebra by subclassing BooleanAlgebra. See also the
tests for other examples of alternative tokenizers.

This tokenizer has these characteristics:

- The `expr` string can span multiple lines,
- Whitespace is not significant.
- The returned position is the starting character offset of a token.
- A TOKEN_SYMBOL is returned for valid identifiers which is a string
  without spaces.

    - These are valid identifiers:
        - Python identifiers.
        - a string even if starting with digits
        - digits (except for 0 and 1).
        - dotted names : foo.bar consist of one token.
        - names with colons: foo:bar consist of one token.
    
    - These are not identifiers:
        - quoted strings.
        - any punctuation which is not an operation

- Recognized operators are (in any upper/lower case combinations):

    - for and:  '*', '&', 'and'
    - for or: '+', '|', 'or'
    - for not: '~', '!', 'not'

- Recognized special symbols are (in any upper/lower case combinations):

    - True symbols: 1 and True
    - False symbols: 0, False and None
z