Optimized contains_prefix

[marvin-j97]

[zaidoon1]

I really want this too! I've asked for it in rocksdb, see: facebook/rocksdb#12396 & facebook/rocksdb#11644 but no movement on that yet.

Also was asked for it by facebook/rocksdb#11899

[marvin-j97]

I think this should be pretty simple, it's more or less just the point read path replaced with a range read instead.

The advantage over tree.prefix().next()?.is_some() is ideally we don't need to seek multiple levels initially to setup the merge iterator. But the worse case cost would probably be the same, unless the prefix does not exist, because then the prefix bloom filters would avoid all I/O (if FPR is small enough).

Prefix bloom filters don't exist yet, but #97

[zaidoon1]

I'll take this one next after the prefix extractor change is merged

[marvin-j97]

I'll take this one next after the prefix extractor change is merged

It may be a while until I get to merging the PR because there's still a bunch of remaining work in 3.0.0 to be done, and I want to reach feature parity to 2.0.0 before I start merging in new features.