"Bitcask can guarantee one disk seek, whereas LevelDB will do one disk seek per level, so at least from that perspective, it can't be better."
Yep, this is a standard tradeoff. When you want your data to be iterable, you have to take the hit. In practice (I oversee a large cassandra cluster), this hit happens about ~1% of the time, which is either a lot, or a little, depending on your constraints.
"Level also has to look down the entire tree if a key is missing."
This is why Cassandra has a bloom filter on top of a very similar data store.
Level also has to look down the entire tree if a key is missing. This means inserts end up being more expensive than reads or updates (which are all just a hash lookup in Bitcask).