Cortex-M4 SIMD

This release adds some Cortex-M4/M7 SIMD instructions to speed up the YCbCr->RGB pixel conversion function. This can speed up overall decode by up to 10%. As of this writing, the code is used on Arduino Nano 33 BLE and the Adafruit SAMD boards, but the build environment of Teensyduino doesn't have the required include files which defined the SIMD macros.