Snapserver responses to return at least one response

[jframe]

PR description

The snap server responses must return at least one response item if the data can be retrieved according to the spec https://github.com/ethereum/devp2p/blob/master/caps/snap.md. This applies to all the request types.

After making this change all the GetByteCodes hive tests are passing. Note still need to make a change to hive testsuite to enable snapserver before this will be seen.

Fixed Issue(s)

Part of #7197

Thanks for sending a pull request! Have you done the following?

• Checked out our contribution guidelines?
• Considered documentation and added the doc-change-required label to this PR if updates are required.
• Considered the changelog and included an update if required.
• For database changes (e.g. KeyValueSegmentIdentifier) considered compatibility and performed forwards and backwards compatibility tests

Locally, you can run these tests to catch failures early:

• unit tests: ./gradlew build
• acceptance tests: ./gradlew acceptanceTest
• integration tests: ./gradlew integrationTest
• reference tests: ./gradlew ethereum:referenceTests:referenceTests

[garyschulte]

LGTM,
one question about the isEmpty() checks, and an ask for comments, but otherwise 🚢

[garyschulte]

sumListBytes operates on the stream, with an orElse(0). it shouldn't matter whether the list is empty or not right? what am I missing here?

[jframe]

The sumListBytes will work fine on an empty list. This change isn't to guard against calling sumListBytes with an empty list but to ensure we only do the sumListBytes(codeBytes) > maxReponseBytes check once we have at least one record in the response. So that we always send at least once record in the response.

[garyschulte]

Is this a "pathological" hive test case where the byteLimit is set too low to allow any records through?

👍 This seems a reasonable fix if the spec says there must always be at least one record in the response, but we should add a comment about why we are doing this, because it isn't obvious why we would do this otherwise.

[jframe]

There wasn't any "pathological" test for the responses that use this predicate, well none obvious anyway. The "pathological" test was for getByteCodes with a limit too small to allow any response. But since the spec does state we need to return at least one response I made the changes anyway to be compliant.