OpenMLDB Benchmark Strategy

[aceforeverd]

First pr #367 , add support for ParameterizedQuery. However, many things left TODO.

Here a some iedas.

• we'd like a comparable graph between OpenMLDB and other memory database e.g memsql, voltdb. https://jmh.morethan.io/ is a nice wheel (and their tips)
  • for every case: move openmldb, memsql and voltdb into single class
  • Don't mix incompatible benchmark styles into one class (like mixing average and single shot is ok, but mixing avarage and throughput doesn't make much sense)
• project structure
  plugins like gradle-jmh-report and jmh-maven-plugin can help order benchmark code.
  • move openmldb-jmh code into openmldb-jdbc instead, inside src/jmh
  • main method in every Benchmark class is not needed
  • and we can add gradle support
• cicd
  • try run benchmark on GitHub workflow, upload test results as json file, so we can visualize it with jmh-visualizer

[aceforeverd]

compare between previous OpenMLDB release might more important at current phase