The benchmarking code is deleted from current master branch. While it's true that the benchmarks haven't kept pace with the rest of the code base, they are still a very valuable tool to quickly check the most used paths(read and write) in KalDB. Instead of deleting the benchmarks, can we update them instead? Some benchmarks are better than no benchmarks.

Without the benchmarks, we may be introducing a lot of regressions that would be extremely hard to catch otherwise.