Update `bench.sh` to use the new `benchmark_runner`

### Is your feature request related to a problem or challenge?

- Part of https://github.com/apache/datafusion/issues/21706
- Follow up to https://github.com/apache/datafusion/issues/21937

#21707 and follow on PRs (e.g. https://github.com/apache/datafusion/pull/23772) added a SQL based benchmark framework with a `benchmark_runner` binary that can run test suites (tpch, clickbench, h2o, imdb, tpcds, etc) either directly or via criterion.

However, `benchmarks/bench.sh` still runs the benchmarks through the old `dfbench` binary, so the two benchmarking paths coexist and we can't remove the old dfbench.

### Describe the solution you'd like

Update `bench.sh` to run benchmarks via the new SQL benchmark framework / `benchmark_runner` for the suites that have been ported, as described in https://github.com/apache/datafusion/issues/21937#issuecomment-4353678551.

### Describe alternatives you've considered

### Additional context

@Omega359 mentioned in https://github.com/apache/datafusion/issues/21937#issuecomment-5196560905 that they planned to file/work on this cleanup.

Start with benchmarks/bench.sh and inspect how it invokes the old dfbench binary. Read the benchmark_runner entry point and the suite mapping in issue #21937's linked comment, then compare the ported suites. Done means bench.sh uses benchmark_runner for every ported suite and no longer depends on the old path for those benchmarks.

Written by the indexing model from the issue text.