For 1.34+: Theoretically, ResourceSlices are just, well, slices of a larger resource pool. Thus, the CapacityPools (and later the Mixins), could be considered unique across the resource pool rather than within the resource slice. If that's true, we could move to a place where we define capacitypools and mixins in one slice, and the devices in different slices. This would allow very large devices partitionable devices that share capacity pools and mixins (think: very large multi-host partitionable devices). Now, we don't do that today because this makes validation hard (impossible?) at admission time. But I want to at least introduce the idea that we might want to do it at some point. Likely we would need a controller to validate across slices and all slices in a pool generation would need to have something set in their status to say they are valid.