RFR: JDK-8328750: [TestBug] Improve Stub Font Support

[Marius Hanl]

On Mon, 25 Mar 2024 23:11:12 GMT, Andy Goryachev <angorya at openjdk.org> wrote:

>> AFAIK, we do not have other tests with that problem.
>> I tried to keep the diff small, but nothing against writing a better method to compare the points here with a delta.
>
> I am not sure why this is needed here, so my concern stems from the fact that `ceil()` might introduce a large bias (0.000001 -> 1.0), and `rnd()` _seems_ to be a better choice.
> 
> But then again, ceil() or rnd() might be a wrong thing to do if we are dealing with scaled coordinates and we don't expect the whole integers, but instead are dealing with small differences between expected and obtained values (in which case a better comparison is required).

I can also use `round` here. Seems to be a bit safer, I agree.

-------------

PR Review Comment: https://git.openjdk.org/jfx/pull/1422#discussion_r1540048636