It's mostly a good thing - but the main problems were still inter-CCD latency (cores on separate CCDs had to talk through the IOD) and that the non-X3D CCD clocked quite a bit higher, so the Windows scheduler just couldn't help but move stuff over to it.
We'll still have the inter-CCD latency issue, but with the Zen 5 X3D CCD in the 9800X3D running almost as fast (and effectively faster) than the 9700X, the solutions to keeping games on a single CCD (preferably the X3D CCD) are potentially as easy as just running the non-X3D CCD slower, if say we got a repeat of the 7900X3D / 7950X3D with only one of two CCDs having stacked cache.
Really, the schedular is just going to have to get significantly smarter. Solutions to making performant but efficient CPUs are unlikely to get less complicated over time, IMO.