Fix secp256k1 verification

[vorot93]

Long map/and_then chains caused the real verification result to be obscured behind layers of Results.

[AgeManning]

This requires storing all underlying values as RLP encoded data.

This looks correct to me as is. After a few more tests, I think we should merge down.

[AgeManning]

This resolves #18

[holiman]

Since this PR fixes #18, it would be good if it also includes the actual ENR-data from #18 as a unit-test too. Personally, I'd import a hundred or so ENR-entries from e.g. https://github.com/ethereum/discv4-dns-lists/blob/master/all.ropsten.ethdisco.net/nodes.json as unit-tests

[AgeManning]

The interesting this is, that one of those ENR's is already a unit test.

The RLP mismatch came up earlier in decoding the actual RLP. @vorot93 found a bug where I missed verifying the signature. Once he corrected that, we realised we were not verifying the signature of this kind of RLP data, because the signature is dependent on how we re-encode the received data.

As a result this PR was failing the unit tests. I corrected this PR to correctly handle this RLP encoding edge case, which in turns lead to passing all the unit tests, which includes an ENR from the list you gave.