bug#56393: Actually fix the long lines display bug

[Gregory Heytings]

> > I can't reproduce either of these two bugs with a regular build, both 
> > of these recipes produce the expected result.  Are these segfaults due 
> > to --enable-checking?
>
> No, I don't think so.  --enable-checking produces SIGABRT, not SIGSEGV.

Yes, I know, but it's possible that some checking causes a segfault.

> The segfault happens because it->bidi_it.bytepos is 1, and FETCH_BYTE 
> cannot be called with a zero argument.  Here's a relevant portion of a 
> debugging session:

Thanks.  With that indication, and after adding assertions in the code, I 
could make Emacs abort when that problem occurred.  I pushed a fix a few 
minutes ago.

But I'm still unable to reproduce your second recipe.  I tried again and 
again, both with a regular and a checking-enable build, with and without 
the fix for the first recipe, and with various frame sizes... but without 
success.  If you can still reproduce it with the fix, could you perhaps 
send me a backtrace?