default pp-fill behaviour

[Madhu]

This is about the state after the series of commits including:

* commit 2f181d60323bd9e0196775828de633100479f4c2
|Author:     Stefan Monnier <monnier@iro.umontreal.ca>
|AuthorDate: Fri Jun 16 13:35:06 2023 -0400
|Commit:     Stefan Monnier <monnier@iro.umontreal.ca>
|CommitDate: Sat Jun 17 17:24:38 2023 -0400
|
|    pp.el (pp-fill): New default pp function
|
|    * lisp/emacs-lisp/pp.el (pp-default-function): Change default.
|    (pp--within-fill-column-p): New helper function.
|    (pp-fill): New function.


The problem is to print an alist in a buffer one element per line:

```
(setq $alist '((1 . a) (2 . b) (3 . c)))

(with-output-to-temp-buffer "*temp*"
  (let ((print-level nil) (print-length nil))
    (pp $alist)))
```

This would give you a temp buffer with this contents

```
((1 . a)
 (2 . b)
 (3 . c))
```

Emacs-30.0.50 now produces

```
((1 . a) (2 . b) (3 . c))
```

If I try to use the new global level variable pp-max-width
pp-use-max-width to prevent filling and get back the old behaviour, I
can't.

```
    (if (fboundp 'pp-28)
        (let ((pp-default-function 'pp-28))
          (pp $alist))
      (pp $alist))))
```

I'd like the old behaviour to be accessible with a global switch.  The
new pp machinery is complex and ( uses the usual impenetrable
undebuggable functional paradims) but if it is not powerful enough
produce the old behaviour in one of its code paths, isn't the power is
in question.