Quantization should preserve sparse accessors

Tested with the sample from https://github.com/donmccurdy/glTF-Transform/issues/289, I'm observing that using quantization on an asset with sparse accessors results in broken morph target animation. For now the quantize function will simply unset the sparse property from any affected accessors. I can't think of any reason that it shouldn't be possible to use the two together, though, so with some more investigation we may be able to fix this.

Also affects Meshopt compression, which begins with a quantization step.