Benchmark runs over time using conbench / ursabench

[alamb]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
We have various benchmarks in datafusion for example Targeted SQL in https://github.com/apache/arrow-datafusion/blob/master/datafusion/core/benches/sql_planner.rs and https://github.com/apache/arrow-datafusion/blob/master/datafusion/core/benches/aggregate_query_sql.rs

There are also some benchmarks for operators that are used as part of other operations, such as https://github.com/apache/arrow-datafusion/blob/master/datafusion/core/benches/merge.rs

We typically run these benchmarks "on demand" but that means we don't have any idea how the benchmark runs vary over time

The rest of the arrow community has a conbench benchmarking tool -- this was introduced to datafusion https://github.com/apache/arrow-datafusion/tree/master/conbench

There are comments on pull requests now like #3062 (comment) that imply we may be able to get runs of the benchmarks on each commit to master and tracked over time.

Describe the solution you'd like
Figure out how to run the datafusion benchmarks regularly and report their results somewhere. Ideally using ursabot / conbench if possible

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered. Ideally w

Additional context
Add any other context or screenshots about the feature request here.

[alamb]

Filed follw on #5504 with more explicit request