```

Benchmark                                             Mode  Cnt  Score   Error  Units
BenchmarkBatchStreamReaders.readLongDecimalNoNull     avgt   60  0.829 ± 0.032   s/op
BenchmarkBatchStreamReaders.readLongDecimalWithNull   avgt   60  0.566 ± 0.015   s/op
BenchmarkBatchStreamReaders.readShortDecimalNoNull    avgt   60  0.544 ± 0.026   s/op
BenchmarkBatchStreamReaders.readShortDecimalWithNull  avgt   60  0.364 ± 0.006   s/op
```

SelectiveStreamReader enables more efficient scan by applying a filter while
extracting values only from the specified rows.

This commit includes an implementation of the new SelectiveStreamReader for the
boolean type.

The new benchmark BenchmarkSelectiveStreamReaders is similar to
BenchmarkBatchStreamReaders, but uses SelectiveStreamReader. The performance
of a single column no filter case is the same for batch and selective readers.

BenchmarkSelectiveStreamReaders.readAllNull                            boolean  avgt   20  0.003 ±  0.001   s/op
BenchmarkSelectiveStreamReaders.readBooleanNoNull                          N/A  avgt   20  0.029 ±  0.001   s/op
BenchmarkSelectiveStreamReaders.readBooleanNoNullWithFilter                N/A  avgt   20  0.098 ±  0.006   s/op
BenchmarkSelectiveStreamReaders.readBooleanWithNull                        N/A  avgt   20  0.098 ±  0.004   s/op
BenchmarkSelectiveStreamReaders.readBooleanWithNullWithFilter              N/A  avgt   20  0.134 ±  0.006   s/op

BenchmarkBatchStreamReaders.readAllNull                  boolean  avgt   20  0.003 ±  0.001   s/op
BenchmarkBatchStreamReaders.readBooleanNoNull                N/A  avgt   20  0.030 ±  0.001   s/op
BenchmarkBatchStreamReaders.readBooleanWithNull              N/A  avgt   20  0.106 ±  0.002   s/op