Remodel boolean queries to hold last query

[valencik]

No description provided.

[valencik]

Benchmarking b2c5a20 with 89c8790 cherry-picked ontop with my AMD Ryzen 7 3800X 8-Core Processor.

Benchmark                                              Mode  Cnt     Score    Error   Units
MapLastTermBenchmark.associativityQueriesMapLastTerm  thrpt   10     8.191 ±  0.171  ops/ms
MapLastTermBenchmark.partialQueriesMapLastTerm        thrpt   10     0.626 ±  0.042  ops/ms
QueryParserBenchmark.associativityQueriesParse        thrpt   10     8.441 ±  0.341  ops/ms
QueryParserBenchmark.orQueries1000Parse               thrpt   10     0.613 ±  0.048  ops/ms
QueryParserBenchmark.orQueries10Parse                 thrpt   10    57.437 ±  0.292  ops/ms
QueryPrinterBenchmark.orQueries1000Print              thrpt   10    28.264 ±  0.396  ops/ms
QueryPrinterBenchmark.orQueries10Print                thrpt   10  4717.728 ± 46.392  ops/ms
QueryPrinterBenchmark.termQueriesPrint                thrpt   10  1082.307 ± 32.293  ops/ms

Compared to "two-nodes" 9efbfed

Benchmark                                              Mode  Cnt     Score     Error   Units
MapLastTermBenchmark.associativityQueriesMapLastTerm  thrpt   10     8.220 ±   0.086  ops/ms
MapLastTermBenchmark.partialQueriesMapLastTerm        thrpt   10     0.592 ±   0.016  ops/ms
QueryParserBenchmark.associativityQueriesParse        thrpt   10     8.396 ±   0.456  ops/ms
QueryParserBenchmark.orQueries1000Parse               thrpt   10     0.608 ±   0.041  ops/ms
QueryParserBenchmark.orQueries10Parse                 thrpt   10    65.001 ±   3.273  ops/ms
QueryPrinterBenchmark.orQueries1000Print              thrpt   10    25.832 ±   4.318  ops/ms
QueryPrinterBenchmark.orQueries10Print                thrpt   10  4424.268 ± 237.150  ops/ms
QueryPrinterBenchmark.termQueriesPrint                thrpt   10  1115.661 ±   7.945  ops/ms

Compared to "main" (from #171 (comment))

Benchmark                                              Mode  Cnt     Score    Error   Units
MapLastTermBenchmark.associativityQueriesMapLastTerm  thrpt   10     8.221 ±  0.234  ops/ms
MapLastTermBenchmark.partialQueriesMapLastTerm        thrpt   10     0.703 ±  0.023  ops/ms
QueryParserBenchmark.associativityQueriesParse        thrpt   10     9.080 ±  0.144  ops/ms
QueryParserBenchmark.orQueries1000Parse               thrpt   10     0.650 ±  0.022  ops/ms
QueryParserBenchmark.orQueries10Parse                 thrpt   10    71.972 ±  2.887  ops/ms
QueryPrinterBenchmark.orQueries1000Print              thrpt   10    28.541 ±  0.392  ops/ms
QueryPrinterBenchmark.orQueries10Print                thrpt   10  4664.162 ± 81.206  ops/ms
QueryPrinterBenchmark.termQueriesPrint                thrpt   10  1091.093 ± 16.845  ops/ms

[valencik]

My current takeaway is that neither the changes here or in #170 are making the parser any faster, just a bit slower.
And while #170 is definitely aimed at making the model more correct, this PR was hopefully going to make mapLastTerm faster. But it's totally dominated by the parsing time anyways.
We also rewrite associateOps in this PR, and it's not immediately obvious that it's a whole lot simpler here.