2.2.0

This was from a code review done by ChatGPT 5.6 Sol.

NvEncSharp 2.2.0 fixes encoding, decoding, color conversion, and native interop issues. (BREAKING CHANGE)

Breaking changes

• Removed NvEncRegisterResourceEx.AsInputPointer() and .AsOutputPointer(). Registered resource handles must be mapped with NvEncoder.MapInputResource() before encoding. Use the returned MappedResource and MappedBufferFmt, then unmap before unregistering. For bitstream output, use NvEncoder.CreateBitstreamBuffer().
• Changed LibCuVideo.GetDecodeStatus()'s output parameter to CuVideoGetDecodeStatus, matching the native API and fixing a potential memory corruption issue. The CuVideoDecoder.GetDecodeStatus() wrapper still returns CuVideoDecodeStatus.

Library fixes

• Fixed generated bitfield accessors, including masks, offsets, and signed values, and regenerated the affected structures.
• Fixed CuVideoFormat.VideoFormat and VideoFullRange bit accessors.
• Corrected NV_ENC_CREATE_INPUT_BUFFER_VER for input buffer creation.
• Fixed missing structure versions in GetEncodePresetConfig(). Marked its overloads obsolete; use GetEncodePresetConfigEx() for P1-P7 presets.
• Fixed NAL packet parsing when data precedes a start code.
• Fixed CUDA color conversion that truncated pixel values, including incorrect RGB24 packing. Updated the bundled native DLL.

Sample fixes

• Fixed screen capture encoding to use mapped resources and release them in the correct order.
• Restored decoded frame copies to host memory.
• Fixed video playback timing and frame queuing, with a 30 FPS fallback when timing metadata is unavailable.
• Fixed buffered NAL submission and end-of-stream flushing.
• Added explicit rejection of unsupported video formats in the decoding sample, which supports 8-bit 4:2:0 H.264 with NV12 output.
• Improved decoding sample shutdown and resource cleanup.

Added Windows x64 native structure size and field offset validation, and expanded NAL parsing tests.