We never render directly to the front buffer, and in case of e.g. a P8 front
buffer, we would fail surface creation if we were to enforce format
restrictions.

wined3d: Use clamp() instead of max() for calculating the clamped cosine in the ffp lighting calculations.

On typical hardware, the clamp() would be an instruction modifier, while the
max() would need an extra instruction.

crypt32: Use CERT_CHAIN_CACHE_ONLY_URL_RETRIEVAL flag for root store certificate's chain validation.

d3dx9: Use the ID3DXInclude object to open the initial file in D3DX[Assemble|Compile|Preprocess]ShaderFromFile.