Issue 5707: fix \cueDuring and \quoteDuring display test

https://sourceforge.net/p/testlilyissues/issues/5707/

These tests are failing in master @ 2ccac23c0e9f7528e55dfd8d711420b4bc5d7a43.
I assume they have been failing for some time.

1: fix \cueDuring and \quoteDuring display test

2: treat warnings as errors in display-lily-tests.ly
so that bugs will not be overlooked.

[Dan Eble, 2020-01-29 00:09:44]

> 2: treat warnings as errors in display-lily-tests.ly
> so that bugs will not be overlooked.

I don't think I tested this part properly.  I'll come back to it.

[dak, 2020-01-29 00:51:52]

On 2020/01/29 00:06:14, Dan Eble wrote:

Pretty sure that the trigger was

commit ed86e71eb0f17e93036142938d8c490a3e2fcbea
Author: Valentin Villenave <valentin@villenave.net>
Date:   Mon Feb 25 17:24:11 2019 +0100

    Use "simple strings" rather than #"hash-prefixed Scheme strings"
    
    Now that the user-exposed syntax hardly requires #-prefixed
    string arguments anymore, we can do away with that syntax
    through various places as well (particularly the LM, where it may
    make things slightly less confusing for newcomers).
    
    This commit leaves LSR-imported snippets untouched for now; it
    does remove a no-longer-relevant sentence in doc-work.itexi,
    and clarifies a few things in fundamental.itely.

and yes, it's not great that this went undetected for that long.  It's not that
the previous output of display-lily was faulty, it's just that this big
replacement changed expectations.

[Dan Eble, 2020-01-29 03:08:48]

change the test, not the display function

[Dan Eble, 2020-01-29 03:12:28]

On 2020/01/29 00:09:44, Dan Eble wrote:
> > 2: treat warnings as errors in display-lily-tests.ly
> > so that bugs will not be overlooked.
> 
> I don't think I tested this part properly.  I'll come back to it.

It works.

[dak, 2020-01-29 06:06:15]

On 2020/01/29 03:08:48, Dan Eble wrote:
> change the test, not the display function

Oh, I am sorry to have been misleading with my comment: I think your original
patch was the right thing to do.  I just remarked that what display-lily printed
would have also worked as input so this was not really bug rather than an uglier
alternative.  But your change of its output clearly was one for the better, even
if the previous output was only uglier than your fix, not strictly wrong.

I should learn to better keep my inclination to lecture in check when there is
nothing to be fixed.

[Dan Eble, 2020-01-29 12:04:56]

revert to original submission

[Dan Eble, 2020-01-29 12:06:58]

On 2020/01/29 06:06:15, dak wrote:
> Oh, I am sorry to have been misleading with my comment

forgiven