[ot][spam][crazy][data] transformer model 'attention' improvement

[Undiscussed Horrific Abuse, One Victim & Survivor of]

-
https://github.com/xloem/transformers/commit/7575b8286dd5c2b328d3c34d9b66dab434282fc0

A draft of calling memory_efficient_attention from the perceiver
model, when configuration parameters are set.
-

Untested. Maybe I can copy google's example again, like before,
somehow, and run the same test with the configuration settings set,
and walk through it to make sure it uses the new code.