Peer's initial hash should use remote_batch to skip files

[elonen]

Currently peer always seems to rehash files on startup. It should probably use remote_batch to skip files that are identical in size and mtime, like it does with local batch after the first hashing, so peer restart would be faster.