CRC issues with igt tests

The kms_color tests expect the default CRC to be downstream of the CTM. Which https://patchwork.freedesktop.org/series/124785/ addressed. But the encoder CRC does not seem to work correctly with the kms_plane test (every CRC readback gets an identical value regardless of what is on screen.

https://patchwork.freedesktop.org/series/125484/ seemed to get the encoder CRC working "correctly" (ie. different images on screen get unique CRC values, which match between sw and hw composition. But that was just based on guessing with an undocumented bitfield, and may not be the correct thing to do.