Quoting the documentation:

tokenize.NL
Token value used to indicate a non-terminating newline. The NEWLINE token indicates the end of a logical line of Python code; NL tokens are generated when a logical line of code is continued over multiple physical lines.

This doesn't seem to be entirely true.
At the end of comments, an .NL, not .NEWLINE is generated:
TokenInfo(type=55 (NL), string='\n', start=(5, 22), end=(5, 23), line='# Some docs for the IF\n')

As a comment cannot extend over several lines, if would appear that a NEWLINE would be appropriate?

See:

https://docs.python.org/3/reference/lexical_analysis.html#blank-lines

OK.
But just to make me understand, would that explain why an else statement gets its DEDENT before the else, instead of before an dedented comment just before it? 
Even if the comment is visually dedented as the else?

Example where tokenize returns the DEDENT in a separate place than where it is visually:

if a=b:
  statement
# Comment
else:
  other statement

In the above case, the DEDENT is reportet before the else, instead of before the comment.

As it says in the docs, a "logical line that contains only spaces, tabs, formfeeds and possibly a comment, is ignored."

If you write:

if a=b:
    statement
# Comment
    another statement

both statements belong to the "if" block.
In your example, the comment is similarly part of the "if" block, so the dedent comes after it.

Ok, I'll work around that then. 
Thanks for your response!