Make DebugLexer.tokenize() more similar to Lexer.tokenize()

[Chris Jerdonek]

Currently, ​DebugLexer.tokenize() looks somewhat dissimilar to ​Lexer.tokenize(), even though they do more or less the same thing. For example, the Lexer class's implementation contains just one call to self.create_token(), whereas DebugLexer contains three calls to it.

This ticket is to make DebugLexer.tokenize() look visibly much more the same as Lexer.tokenize(), and in particular simpler. One advantage is that it will be more obvious on inspection that DebugLexer acts the same as Lexer. Another advantage is that it will be easier to keep the two implementations in step when applying optimizations to Lexer. It will be more maintainable and less likely to introduce bugs if the implementations for the two lexers don't diverge too much.

The idea will become more clear once the PR is posted.

[Chris Jerdonek]

PR: ​https://github.com/django/django/pull/14753

[Mariusz Felisiak <felisiak.mariusz@…>]

In b2be7e1:

Refs #33002 -- Made template_tests.tests.py's tests test both Lexer and DebugLexer.

[Mariusz Felisiak <felisiak.mariusz@…>]

In 65ed96fa:

Refs #33002 -- Added DebugLexer/Lexer.tokenize() test for trailing text token.

[Mariusz Felisiak <felisiak.mariusz@…>]

In 6242c22:

Refs #33002 -- Optimized Lexer.tokenize() by skipping computing lineno when not needed.

[Mariusz Felisiak <felisiak.mariusz@…>]

In 6fedd868:

Refs #33002 -- Renamed variable from bit to token_string in Lexer.tokenize().

[Mariusz Felisiak <felisiak.mariusz@…>]

In f0776a55:

Fixed #33002 -- Made DebugLexer.tokenize() more closely parallel Lexer.tokenize().