Add geometric mean normalization for scores

[martin-gaievski]

Description

Adding geometric mean technique, that is a generalization of the mean that is based on product and N-th root of N values (more details here). Weights are supported similarly what it's done in arithmetic mean. Example of pipeline with processor config:

{
    "description": "Post processor for hybrid search",
    "phase_results_processors": [
        {
            "normalization-processor": {}, 
            "combination": {
                 "technique": "geometric_mean",
                  "parameters": {
                        "weights": [
                            0.4, 0.7
                        ]
                    }
             }
        }
    ]
}

In addition to main changes there are some refactoring in integ tests. I have to put it to this PR because with few new tests added for geometric mean auto redeploy feature started acting more aggressively and tests became flaky.

• move from global model id to reading it at the beginning of each test case. this is required because model can be redeployed by the auto redeploy feature of ml-commons
• split on large class into three smaller ones based on area of responsibility
• cleanup outdated tests

Issues Resolved

#228, part of solution for #126

Check List

• New functionality includes testing.
  • All tests pass
• New functionality has javadoc added
• Commits are signed as per the DCO using --signoff

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[codecov]

Codecov Report

Merging #239 (f2fdcbe) into feature/normalization (6ad641a) will increase coverage by 3.80%.
Report is 1 commits behind head on feature/normalization.
The diff coverage is 93.93%.

@@                     Coverage Diff                     @@
##             feature/normalization     #239      +/-   ##
===========================================================
+ Coverage                    82.43%   86.23%   +3.80%     
- Complexity                     323      337      +14     
===========================================================
  Files                           26       28       +2     
  Lines                          979      981       +2     
  Branches                       153      153              
===========================================================
+ Hits                           807      846      +39     
+ Misses                         108       69      -39     
- Partials                        64       66       +2     

Files Changed	Coverage Δ
...ation/ArithmeticMeanScoreCombinationTechnique.java	89.47% <88.88%> (-2.84%)	⬇️
...ination/HarmonicMeanScoreCombinationTechnique.java	94.11% <91.66%> (+94.11%)	⬆️
...nation/GeometricMeanScoreCombinationTechnique.java	94.11% <94.11%> (ø)
...ch/processor/combination/ScoreCombinationUtil.java	95.45% <95.45%> (ø)
...processor/combination/ScoreCombinationFactory.java	100.00% <100.00%> (ø)

[heemin32]

I still have a doubt on the effectiveness of this test code. l believe we don’t need test code based on random number. I would like to hear other opinions though.

[navneet1v]

I don't have any concern as long as test is able to fail if there are changes in formula. Let's just make sure that it doesn't become flaky because of floating point precision losses.