Implemented experimental HLSL post-processing support on Direct3D 9
targets. [Ryan Holtz, Bat Country Entertainment]


 * 5-pass post-processing: Upscale, Post-Process, Store Last Frame,
    Defocus 1, Defocus 2
 * Many tunable effects including: Scanlines, defocus, linear
    deconvergence, radial deconvergence, pincushion, RGB colorspace
    convolution, YIQ colorspace convolution, saturation, simulated dot
    crawl, simulated chroma subsampling, aperture masking, and more.
 * Requires a GPU that supports Shader Model 3.0 to be enabled and a
    powerful GPU, the entire pipeline consists of approximately 30
    texel fetches and approximately 230 arthimetic ops.
 * Will supersample the framebuffer up to 9x in both X and Y, but this
    requires an enormously powerful GPU that has not been invented;
    users with Radeon 5000-class cards should limit themselves to 3x,
    Radeon 4000 to 1.5x.
 * The default configuration will NOT appear to do anything; it
    requires tuning to the user's liking.
 * Should nicely fall back in all cases except missing shaders, and it
    might fall back correctly in that case as well. Report any
    anomalies.
- For obvious reasons, the Direct3D8 renderer cannont support this.

HLSL Updates [Ryan Holtz, Bat Country Entertainment]
 * Split color convolution and deconvergence into separate shaders for
    potential GPU savings down the line
 * Added light and heavy variants of the color convolution shader, the
    former with YIQ colorspace removed
 * Re-worked defocus to occur prior to shadow mask application, as it
    would be on a real monitor.
 * Split "pincushion" into "Pincushion" and "Screen Curvature", the
    former affecting the only the displayed image and the latter only
    affecting the shadow mask.
 * Made major fixes to CVBS simulation, significantly increasing color
    saturation
 * Added magnet deconvergence via a shader uniform
 * Improved phosphorescence
 * Improved YIQ codec
 * Added adjustable color carrier in YIQ processing mode. 
 * Fixed resolution-change crash and likely D3D performance regression
    with -nohlsl on low-spec cards
 * Created a new OSD function, osd_get_slider_list, which allows
    OS-specific slider controls.

HLSL Post-Processing Updates: [Ryan Holtz, Bat Country Entertainment, cgwg]
 * The defocus pass is now switched off when defocus_x and defocus_y
    are zero, allowing finer-grained performance tuning.
 * Removed YIQ convolution from the main color-convolution shader and
    replaced it with a full composite encode/decode pass. This is slower,
    but looks amazing(ly like a terrible TV) and can be turned off.
 * More authentic NTSC dot crawl and bandwidth limiting.