We should not be running pkh logic through rawpkh

[apoelstra]

See for example https://github.com/rust-bitcoin/rust-miniscript/blob/master/src/miniscript/satisfy.rs#L937 (not sure if there are any other examples). Here we have a public key, convert it to a hash, then look up the key again from that.

In #481 (the "planning module" PR) I am suggesting that we work with DescriptorPublicKeys rather than DefinitePublicKeys, which means that we may be working with keys where we don't know their hash160 and can't do this lookup. So in addition to being unnecessary, this roundtrip through a hash may become wrong.

[sanket1729]

So in addition to being unnecessary, this roundtrip through a hash may become wrong.

I agree this is unnecessary, but why can it be wrong?

[apoelstra]

@sanket1729 imagine we had a satisfier that simply did not have a map from hashes to public keys (because it assumes we have no raw pkhs), so the hash lookup would always return None.