bug#63790: 30.0.50; prog-fill-reindent-defun regression

[Juri Linkov <juri@linkov.net>]

>> This regression is in master, not in emacs-29.
>
> If it is a regression, then compared to what? emacs-29 doesn't have this
> function. Compared to some earlier revision?

Sorry, the subject was not precise.  This is more elaborate:
the new function 'prog-fill-reindent-defun' caused a regression for
'M-q' compared to emacs-29 where 'M-q' was bound to 'fill-paragraph'.

>> 0. emacs-30 -Q
>> 1. add to the beginning of the *scratch* buffer a list, so that
>> *scratch* looks like this:
>> (+
>>   1
>>      2
>>   3)
>> ;; This buffer is for text that is not saved, and for Lisp evaluation.
>> ;; To create a file, visit it with C-x C-f and enter text in its buffer.
>> 2. Activate the region with the beginning at the start of the line with
>> the first comment, and the region end with point at the end of the buffer.
>> 3. Type 'M-q' (prog-fill-reindent-defun)
>> It indents the list instead of the comment.
>> But when point is at the region beginning then 'M-q' correctly indents
>> the comments.
>
> This happens because in this scenario point ends up outside of the comment
> (at eob). So when the function is called, in looks for a defun.
>
> Did we at some point add (or decide to add) a condition when, if a region
> is active, it should only refill and not reindent?

Maybe the logic of region detecting/handling could be copied from
'fill-paragraph' to 'prog-fill-reindent-defun'?