RFR: 7902808: Batch size tests still fail due to numerical and timing issues

[Aleksey Shipilev]

Current batch size tests are still flaky, experiencing intermittent failures in GH actions. It seems to be caused by individual timing measurements per @Benchmark. If test are stalled outside of that measurement, the sum of individual measurements can be much lower than the actual wall clock time.

But what's significantly worse, the operation handling *divides* real operation count by batch size, which is subject to gross rounding errors, as the division happens in long domain. Changing that to double should be enough to regain the accuracy.

-------------

Commit messages:
 - 7902808: Batch size tests still fail due to numerical and timing issues

Changes: https://git.openjdk.java.net/jmh/pull/13/files
 Webrev: https://webrevs.openjdk.java.net/?repo=jmh&pr=13&range=00
  Issue: https://bugs.openjdk.java.net/browse/CODETOOLS-7902808
  Stats: 79 lines in 11 files changed: 40 ins; 6 del; 33 mod
  Patch: https://git.openjdk.java.net/jmh/pull/13.diff
  Fetch: git fetch https://git.openjdk.java.net/jmh pull/13/head:pull/13

PR: https://git.openjdk.java.net/jmh/pull/13