1. Changing extension of files with binary data to .bin.
2. Running separate benchmark for each query according to numElements in their
result.

1. Exporting query and result byte slices inside SubGraph so that they are
written to file when we write the SubGraph to file after gobencoding it.
2. Added benchmark test to benchmark ToJSON and ToProtocolBuffer methods for
actors and directors query for 10,100 and 1000 entities in the result.

preTraverse()

Removed the unnecessary extract function and result struct. Now preTraverse
directly acceses flatbuffers to get values and uidmatrix.
Renamed struct fields in the protocol buffer schema file.

1. Removed client.go and client_test.go
2. Modified protocol buffer structure to have Values as a struct which
contains byte slice.
3. Added support to calculate latency because of protocol buffer.

1. Protocol buffer now stores a entity based subgraph instead of a predicate
based subgraph.

Query would be needed to parse the Graph Response on the client side, hence
sending it in the protocol buffer response.

1. Renamed GetResponse to Query in DGraph service.
2. Client now accepts the ip of the server as a flag instead of just the port.
3. Removed ToProtocolBuffer method, and added PreTraverse as a method for a Subgraph.
4. Renamed runServerForClient to runGrpcServer.

1. Added subgraph.proto file which has the message structure.
2. Added ToProtocolBuffer method which converts a subgraph to protocol buffer.

Remove gql module's dependency on query. Make it the other way round. GraphQL query string, now gets converted into query.SubGraph as an additional step run via query.ToSubGraph.

Use cache for streaming entries instead of opening files for each posting list. Use iterator instead of channel while streaming entries.

Loads of fixes and tweaks to make data loading work. Largely, the system is really fast in inserting data, but memory becomes a problem. So, running continuous memory checks, and if the db exceeds set memory, then stopping in world in batch load mode to deal with it first. In server mode, same code runs but without stopping the world.

Make list matrix work, yeah! This would serialize uids for the next query, which would then again return a matrix, so on and so forth.