[GPU] Mostly generic, not square-only resolution scaling

src/xenia/gpu/shaders/host_depth_store_4xmsaa.cs.hlsl

@@ -6,20 +6,17 @@ Texture2DMS<float> xe_host_depth_store_source : register(t0);
 [numthreads(8, 8, 1)]
 void main(uint3 xe_thread_id : SV_DispatchThreadID) {
   // 1 thread = 8 samples (4x0.5 pixels, resolve granularity is 8 pixels).
-  uint resolution_scale = XeHostDepthStoreResolutionScale();
-  // Group height is aligned to resolve granularity, Y overflow check not
-  // needed.
-  [branch] if ((xe_thread_id.x >> 1u) >
-               XeHostDepthStoreWidthDiv8Minus1() * resolution_scale) {
+  // Group height can't cross resolve granularity, Y overflow check not needed.
+  [branch] if ((xe_thread_id.x >> 1u) >= XeHostDepthStoreScaledWidthDiv8()) {
     return;
   }
-  uint2 pixel_index = XeHostDepthStoreOrigin() * resolution_scale +
+  uint2 pixel_index = XeHostDepthStoreScaledOrigin() +
                       uint2(xe_thread_id.x << 2u, xe_thread_id.y >> 1u);
   // For simplicity, passing samples directly, not pixels, to XeEdramOffsetInts.
   uint edram_address_int4s =
       XeEdramOffsetInts((pixel_index << 1u) | (xe_thread_id.xy & 1u), 0u,
                         XeHostDepthStorePitchTiles(), kXenosMsaaSamples_1X,
-                        false, 0u, 0u, resolution_scale)
+                        false, 0u, 0u, XeHostDepthStoreResolutionScale())
       >> 2u;
   // Render target horizontal sample in bit 0, vertical sample in bit 1.
   int source_sample_left = int((xe_thread_id.y & 1u) << 1u);