move sha256 coprocessor out of lurk crate

[porcuquine]

The sha256 coprocessor does not belong in the Lurk crate. I think it was moved there as part of an attempt to increase code sharing in examples and benches that use it. It would be better to solve the problem through some other technique.

If Lurk had some need for a truncated sha256 operation, or was highly opinionated about wanting to provide this to others, then maybe it would make sense. As it stands, we're forced to concern ourselves with whether that coprocessor is correct, or what it would even mean for it to be correct. It's not that these are entirely uninteresting questions, but they cost more than they are worth. If this code did not live in the Lurk crate, it would be easier for us to formally declare our appropriate apathy.

[arthurpaulino]

More context: #732 did this because the sha256 coprocessor code was replicated in three different places (benchmark + ivc example + nivc example). Now that it's centralized, it's easier for us to move it elsewhere