Dual JVM benchmark computing mean on a table loaded by Generic_JDBC driver

[JaroslavTulach]

Pull Request Description

• dual JVM benchmark to track progress of Use Arrow format for LongStorage to make exchange between JVMs and Python possible #13851
• Column is loaded by Standard.Generic_JDBC module, but the mean is computed by Standard.Table
• that (in dual JVM mode) means the Column data are crossing the boundary

Important Notes

• with c35aa2e we can run both tests at the same time:

sbt:enso> std-benchmarks/benchOnly JDBC.mean
[info] Benchmark                     Mode  Cnt    Score   Error  Units
[info] Dual_JVM_Generic_JDBC.mean    avgt       544,708          ms/op
[info] Single_JVM_Generic_JDBC.mean  avgt        32,655          ms/op

• e.g. 20x slower for now
  • Arrow like storage over ByteBuffer to allow dual JVM buffer exchange #14309 will make it faster

Checklist

Please ensure that the following checklist has been satisfied before submitting the PR:

• All code follows the
  Scala,
  Java,
• Benchmark is executed in single as well as dual JVM mode during our CI run

[Akirathan]

In the future, it would be nice if the benchmark would also ensure that the Column class was really loaded by the other VM. This check could be put next to the Properly computed mean assert. Currently, the only way to know that the Column is really exchanged between two VMs is that the other bgenchmark is slower.