8bit encodes with HIGH_BIT_DEPTH builds

[Steve Borho]

We either need to fix the new assembly routines to properly support 8bpp encodes with HIGH_BIT_DEPTH builds or we need to add checks to explicitly prevent this configuration at runtime. Today, it just outputs garbage streams which is unacceptable.

[Steve Borho]

Feedback from a consortium member:

Yep, having to have three versions of the static library for 8bit, 10 bit and 12 bit encodes would be problematic for us. So in order of preference,

1) It's all handled in a single static library with the different code pieces switched out by you :-) 2) Different namespaces for different bit depths 3) The DLL option

[Steve Borho]

when the input pictures have bit depth greater than 8, we should mask input pixels to ensure we do not exceed dynamic range anywhere.

[Steve Borho]

Note: we currently prevent 8bit encodes with 16bpp builds

[Steve Borho]

bit depth is now always a compile time decision, finalized by 7eadf46a7e04