Slight mask optimization

[kgrabs]

For L134 it's a little faster to use the built-in parameter scale=2. The results are not exact but are ±1 with integer formats so that can be chalked up to rounding errors.

A quick test with ffms2 video input gave me

• 264fps vs 285fps in 8 bit
• 211fps vs 241fps in 16 bit
• 150fps vs 191fps in 32 bit

[DJATOM]

You can put external whateveryouwant mask clip and it will have zero overhead against internal one.
The bottleneck of this script is not a mask anyway, but eedi3.

[kgrabs]

Yeah that's more or less why I didn't bother to make it a PR, but it wouldn't hurt to add it in next time there's a meaningful update.

[DJATOM]

I'll probably break things once again and allow to pass kwargs into filters in the next updates.