Add сontrols for colour format/range conversion

[fifonik]

[ls-milkyway]

Well....I request you not to change any parameters of a distorted file ...using FFmpeg (except scaling ...that too if user has not scaled it ....and always use BICUBIC method for scaling it to reference size...as per VMAF recommendations)... Reasons:

1. Distorted file is the result of an encode process ....to compare ...it is the encoders job to suit it to be used for PSNR and/or SSIM and/or VMAF...it's not your task ....max you can do is: to scale distorted file to reference file for VMAF .....that too using BICUBIC only....this way you will not hamper quality of the distorted file much.

2. There are other programs in python that too derive PSNR,SSIM, & VMAF etc. & if users find that they are showing better results than FFmetrics (due to excessive changes made by you on the distorted file....liker color format, range etc.....unless you can do that without re-encoding the distorted file...) then they will switch to python solutions.

3. This can also induce false sense in the users ....due to low scores....that encoding further needs improvement in cases when max tuning is already implemented...in other words ...less you touch distorted...better the image of the results user will get.

4. My encoding experience says never use FFmpeg or hardware encoders for encoding purpose.....you will always get poor results....always use a good CPU (with supporting instruction sets) for encoding with software based on x264, x265 or other highly efficient encoders like av1 (if compatibility is not a problem)....

This discussion also answers some of the queries at #58