Performance regression after reordering optimizer pipeline

[hmottestad]

GH-3969 moves the filter optimizer causing a performance regersion.

[hmottestad]

Current (main branch)

Benchmark                                                     Mode  Cnt     Score   Error  Units
QueryBenchmark.complexQuery                                   avgt          1.532          ms/op
QueryBenchmark.different_datasets_with_similar_distributions  avgt       3620.350          ms/op  <---
QueryBenchmark.groupByQuery                                   avgt          0.724          ms/op
QueryBenchmark.long_chain                                     avgt        271.970          ms/op
QueryBenchmark.lots_of_optional                               avgt         53.329          ms/op
QueryBenchmark.minus                                          avgt       2437.301          ms/op
QueryBenchmark.nested_optionals                               avgt         75.727          ms/op
QueryBenchmark.pathExpressionQuery1                           avgt          7.214          ms/op
QueryBenchmark.pathExpressionQuery2                           avgt          1.137          ms/op
QueryBenchmark.query_distinct_predicates                      avgt         62.587          ms/op
QueryBenchmark.simple_filter_not                              avgt          3.138          ms/op

[hmottestad]

After reverting the order of the optimizer pipeline

Benchmark                                                     Mode  Cnt     Score   Error  Units
QueryBenchmark.complexQuery                                   avgt          1.454          ms/op
QueryBenchmark.different_datasets_with_similar_distributions  avgt          0.724          ms/op  <---
QueryBenchmark.groupByQuery                                   avgt          0.853          ms/op
QueryBenchmark.long_chain                                     avgt        263.109          ms/op
QueryBenchmark.lots_of_optional                               avgt         54.652          ms/op
QueryBenchmark.minus                                          avgt       2432.936          ms/op
QueryBenchmark.nested_optionals                               avgt         73.267          ms/op
QueryBenchmark.pathExpressionQuery1                           avgt          7.990          ms/op
QueryBenchmark.pathExpressionQuery2                           avgt          1.122          ms/op
QueryBenchmark.query_distinct_predicates                      avgt         63.506          ms/op
QueryBenchmark.simple_filter_not                              avgt          2.877          ms/op

[hmottestad]

@jeenbroekstra moving the filter optimizer before the join optimizer had some unforseen performance issues with one of the benchmarks. I'm going to revert the change for now.