dag/resolve: should return without loading the block for empty paths

[koivunej]

This affects #236. Having previously gone back and forth with the "initial block" blocking I wonder if this is something which was changed recently in the conformance tests.

[koivunej]

Wondering if this should affect how the "the upcoming" ipfs::Ipfs::resolve or at the moment ipfs_http::v0::refs::walk_path should work, or if this should be just hacked in dag/resolve. Wrote #238 on "the upcoming ...::resolve".

[koivunej]

I was expecting this "bug" to be hit during #229 however it wasn't, so I must have misunderstood the bug cause. Good if this doesn't need to be touched at least for now (I've danced with this perhaps in #184).