feat: implement production-ready vector storage

[kp2pml30]

kd tree modification (with balanced insertion) or other

It needs to have:

1. fast insertion/deletion (o(N) ; small o)
2. limitations:
   1. Can't use random (no random in det mode)
   2. Can't use amortized operations (transaction cost must be predictable)
   3. Can't use background rebuilding (we are on a blockchain)
3. it must use our storage types (TreeMap, DynArray)
4. k << n, not constant
5. it would be best if we could use arbitrary distance, but it is not a must

[kp2pml30]

PH-tree: issue: duplicate keys are not allowed

[kp2pml30]

R-tree: todo research

R-trees do not guarantee good worst-case performance, but generally perform well with real-world data. While more of theoretical interest, the (bulk-loaded) Priority R-tree variant of the R-tree is worst-case optimal, but due to the increased complexity, has not received much attention in practical applications so far.

[kp2pml30]

Our limitations:

1. Can't use random
2. Can't use amortized operations
3. Can't use background rebuilding