Speedup GlobalHitsThresholdChecker a little

[original-brownbear]

Even though this field is not volatile, writing it isn't free and causes needless cache thrashing at some frequency. We can speed things up by only writing the true value and never the false value.

At least on x86 this has a small but clearly visible positive impact in wiki-med-all benchmarks:

                            TaskQPS baseline      StdDevQPS my_modified_version      StdDev                Pct diff p-value
                      AndHighLow     1320.25      (7.8%)     1303.53     (13.6%)   -1.3% ( -21% -   21%) 0.718
                        HighTerm      855.19      (6.9%)      848.11     (14.1%)   -0.8% ( -20% -   21%) 0.813
         AndHighMedDayTaxoFacets      134.18      (3.8%)      133.19      (4.5%)   -0.7% (  -8% -    7%) 0.580
          OrHighMedDayTaxoFacets        6.73      (5.2%)        6.73      (4.1%)   -0.0% (  -8% -    9%) 0.982
            MedTermDayTaxoFacets       13.45      (3.9%)       13.45      (5.3%)    0.0% (  -8% -    9%) 0.982
           BrowseMonthSSDVFacets        4.49      (8.2%)        4.51      (8.7%)    0.3% ( -15% -   18%) 0.911
                 MedSloppyPhrase       36.33      (7.0%)       36.45      (6.0%)    0.3% ( -11% -   14%) 0.873
                       OrHighMed      224.17      (3.8%)      225.47      (4.9%)    0.6% (  -7% -    9%) 0.675
                     MedSpanNear       96.02      (7.3%)       96.60      (7.2%)    0.6% ( -12% -   16%) 0.790
     BrowseRandomLabelSSDVFacets        3.26      (2.3%)        3.29      (2.6%)    0.9% (  -3% -    5%) 0.227
                    OrNotHighLow     1097.11     (11.4%)     1107.99      (9.6%)    1.0% ( -17% -   24%) 0.766
       BrowseDayOfYearSSDVFacets        4.38      (3.4%)        4.42      (3.9%)    1.0% (  -6% -    8%) 0.377
            HighTermTitleBDVSort       24.30      (7.0%)       24.61      (8.0%)    1.3% ( -12% -   17%) 0.598
                         Prefix3      102.78      (8.3%)      104.17      (8.1%)    1.4% ( -13% -   19%) 0.603
                          Fuzzy1       91.69      (6.5%)       92.98      (5.9%)    1.4% ( -10% -   14%) 0.469
                       LowPhrase       71.06      (5.6%)       72.07      (4.0%)    1.4% (  -7% -   11%) 0.355
                        PKLookup      233.39      (3.5%)      237.05      (3.0%)    1.6% (  -4% -    8%) 0.130
                    HighSpanNear        2.51      (2.2%)        2.56      (2.8%)    1.8% (  -3% -    6%) 0.028
                     AndHighHigh      115.40      (3.9%)      117.54      (3.1%)    1.9% (  -4% -    9%) 0.096
            HighIntervalsOrdered       22.18      (9.8%)       22.62      (7.5%)    2.0% ( -13% -   21%) 0.471
                          IntNRQ      832.49     (11.0%)      849.89     (15.9%)    2.1% ( -22% -   32%) 0.628
                       MedPhrase      327.84      (5.9%)      335.15      (8.0%)    2.2% ( -11% -   17%) 0.316
            BrowseDateSSDVFacets        1.24      (3.4%)        1.27      (4.1%)    2.3% (  -4% -   10%) 0.048
                         Respell       44.23      (2.5%)       45.34      (2.2%)    2.5% (  -2% -    7%) 0.001
                 LowSloppyPhrase       31.93      (8.7%)       32.74      (8.4%)    2.5% ( -13% -   21%) 0.348
                      OrHighHigh       74.05      (6.3%)       76.04      (7.4%)    2.7% ( -10% -   17%) 0.216
                HighSloppyPhrase       33.90     (11.0%)       34.87      (8.9%)    2.8% ( -15% -   25%) 0.368
             MedIntervalsOrdered       27.04      (3.9%)       27.82      (4.7%)    2.9% (  -5% -   12%) 0.034
                    OrHighNotMed      468.83      (6.9%)      484.01      (8.1%)    3.2% ( -11% -   19%) 0.175
        AndHighHighDayTaxoFacets       23.97      (5.4%)       24.77      (4.7%)    3.3% (  -6% -   14%) 0.039
                          Fuzzy2       81.16      (3.1%)       84.04      (3.0%)    3.6% (  -2% -   10%) 0.000
                     LowSpanNear       47.37      (6.6%)       49.13      (4.8%)    3.7% (  -7% -   16%) 0.041
                        Wildcard      193.09     (11.7%)      200.32     (11.6%)    3.7% ( -17% -   30%) 0.309
                      AndHighMed      253.31      (8.0%)      263.49      (5.9%)    4.0% (  -9% -   19%) 0.072
                         LowTerm      586.07      (7.0%)      609.94     (11.3%)    4.1% ( -13% -   24%) 0.171
               HighTermTitleSort       59.11      (6.3%)       61.53      (5.3%)    4.1% (  -7% -   16%) 0.026
                   OrNotHighHigh      323.26      (6.3%)      337.69      (8.2%)    4.5% (  -9% -   20%) 0.055
            BrowseDateTaxoFacets        4.89      (5.3%)        5.11      (6.3%)    4.6% (  -6% -   17%) 0.012
             LowIntervalsOrdered       53.41      (4.9%)       55.90      (5.3%)    4.7% (  -5% -   15%) 0.004
     BrowseRandomLabelTaxoFacets        4.24      (4.1%)        4.44      (3.9%)    4.8% (  -3% -   13%) 0.000
                   OrHighNotHigh      509.58      (8.2%)      534.81      (6.3%)    5.0% (  -8% -   21%) 0.032
                    OrNotHighMed      356.67      (4.8%)      374.35      (6.3%)    5.0% (  -5% -   16%) 0.005
               HighTermMonthSort     1286.76     (10.8%)     1350.90     (11.4%)    5.0% ( -15% -   30%) 0.155
                      TermDTSort      194.58      (6.9%)      204.67      (8.6%)    5.2% (  -9% -   22%) 0.036
                      HighPhrase       19.20      (7.9%)       20.21      (8.0%)    5.2% (  -9% -   22%) 0.038
           HighTermDayOfYearSort      330.05     (11.2%)      347.39     (11.2%)    5.3% ( -15% -   31%) 0.139
                       OrHighLow      447.18      (6.8%)      470.74      (7.2%)    5.3% (  -8% -   20%) 0.017
                         MedTerm      691.26      (7.0%)      727.87      (9.0%)    5.3% (  -9% -   22%) 0.037
                    OrHighNotLow      396.54      (6.5%)      419.54      (9.7%)    5.8% (  -9% -   23%) 0.027
       BrowseDayOfYearTaxoFacets        4.95      (4.7%)        5.25      (6.6%)    5.9% (  -5% -   18%) 0.001
           BrowseMonthTaxoFacets       12.00     (31.4%)       12.91     (29.4%)    7.6% ( -40% -   99%) 0.429

[javanna]

LGTM

[original-brownbear]

Thanks Luca!