Hardware support can always be better, naturally!
That being said, we need to be clear here - it is not a matter of LMCE supporting hardware, it is more a matter of hardware supporting LMCE (or more specifically Linux and certain OpenGL requirements). LMCE will work with any GPU hardware as long as the manufacturer has decent support of the Linux platform and a reasonable OpenGL implementation.
Unfortuantely, only nVidia has bothered with this so far. Their driver is pretty good, but still has big holes - but most importantly it is not open source, nor is the "API" for accessing their hardware... so when there is a bug or a new feature required, we have to wait for nVidia to fix/implement it. This is very poor form, but still light years ahead of most of the other manufacturers. That being said, Intel and ATI seem to be moving towards a more supported and open model, so this may change in the medium term.
But for now, you have the selection of pretty much
any GPU from probably the biggest GPU manufacturer. That's hardly a limitation
And you
can get Via, Intel and ATI hardware to work as long as you use UI1.
As for the 9200 - clearly it isn't as fast as higher models in their range, and that is the point, really. However, it is vastly faster at OpenGL manipulation and rendering than anything that LMCE is likely to need, and I believe that the video decode acceleration is probably more than fast enough for 1080p - 264, so it really doesn't matter all that much... maybe even save you money!
If you have the link to the comparison (particularly video decode acceleration) please post it so that we can determine where the chip stands in performance.
As far as Microsoft is concerned, the corollary of the above is true too.... it isn't that Microsoft supports more hardware, it is that more hardware supports Microsoft, for the obvious reasons. An academic point, maybe, but it is important to realise that support says nothing about the relative qualities of either Windows or Linux MCE.
