Optimize the join operators

DataFusion is underperforming the Polars streaming engine on some localhost join queries (1e8 rows of data on a Macbook M3 with 16GB of RAM):

Here are the [join queries](https://github.com/apache/datafusion/blob/main/benchmarks/queries/h2o/join.sql).

I am guessing the join operator can be optimized, similar to how the filtering and aggregation operations were optimized.

Here is an example of how the median function was made faster: https://github.com/apache/datafusion/issues/13550

See this epic for more info: https://github.com/apache/datafusion/issues/13548

Start with the H2O join queries in benchmarks/queries/h2o/join.sql and the linked epic for context on the performance work. The issue does not identify a specific join operator or optimization, so first investigate which queries lag and compare with the approach in issue 13550. Done would mean measurable improvement on the reported join workload; no target is specified.

Written by the indexing model from the issue text.