Pushed to my private branch.

https://git.savannah.gnu.org/cgit/groff.git/commit/?h=branden-2023-05-22&id=6a2e3a59087c103524dd3410c3f148b506c11bbc

Affects groff 1.22.4.

It seems to be a common error in *roff to use `ne` without `br` before it, but the behavior you get when including the `br` request is usually what people want.

$ cat EXPERIMENTS/use-ne-to-break-page.roff
.de LI
Sed ut perspiciatis, unde omnis iste natus error sit voluptatem
accusantium doloremque laudantium, totam rem aperiam eaque ipsa, quae ab
illo inventore veritatis et quasi architecto beatae vitae dicta sunt,
explicabo.
..
.LI
.ne 999i
.br
Now check this out.
.sp
.LI
Sed ut perspiciatis, unde omnis iste natus error sit voluptatem
accusantium doloremque laudantium, totam rem aperiam eaque ipsa, quae ab
illo inventore veritatis et quasi architecto beatae vitae dicta sunt,
explicabo.
.br
.ne 999i
$ nroff EXPERIMENTS/use-ne-to-break-page.roff | cat -s
Sed  ut  perspiciatis, unde omnis iste natus error sit voluptatem
accusantium doloremque laudantium, totam rem aperiam eaque  ipsa,
quae ab illo inventore veritatis et quasi architecto beatae vitae

dicta sunt, explicabo.
Now check this out.

Sed ut perspiciatis, unde omnis iste natus error  sit  voluptatem
accusantium  doloremque laudantium, totam rem aperiam eaque ipsa,
quae ab illo inventore veritatis et quasi architecto beatae vitae
dicta  sunt, explicabo.  Sed ut perspiciatis, unde omnis iste na‐
tus error sit voluptatem accusantium doloremque laudantium, totam
rem aperiam eaque ipsa, quae ab illo inventore veritatis et quasi
architecto beatae vitae dicta sunt, explicabo.

I propose a simple patch for this.

diff --git a/contrib/mm/m.tmac b/contrib/mm/m.tmac
index c679a7089..0b2878c89 100644
--- a/contrib/mm/m.tmac
+++ b/contrib/mm/m.tmac
@@ -1297,6 +1297,7 @@ numeric; got '\\$1'
 .\" stop output and begin on next page. Fix footnotes and all that.
 .de pg@next-page
 .\".debug next-page
+.br
 .ne 999i               \" activate trap
 .\" .pg@footer
 ..

Unfortunately, `pg@next-page` has several callers and it's conceivable that some of them might exhibit changed behavior due to this.  So make this change in the 1.24 cycle and see what shakes out.

Also, with any luck we'll have the new, improved `.R` register by then and we can use that as the argument to `ne` instead of `999i`, which risks numeric overflow in the (admittedly unlikely) event that some output device shows up with a resolution so high that this value causes numeric overflow.  See bug #63587.