Support benchmarking of Solr streaming expressions

## Summary
Solr [Streaming Expressions](https://solr.apache.org/guide/solr/latest/query-guide/streaming-expressions.html)
are a powerful distributed computation API, but Solr Orbit has no dedicated support for
benchmarking them. This issue tracks adding first-class streaming expression support.

## Motivation
- Streaming expressions cover aggregation, faceting, parallel SQL, graph traversal, and
machine learning scoring — workloads that differ significantly from standard query
benchmarks
- Teams running Solr for analytics or ETL pipelines need a way to compare expression
performance across Solr versions or cluster configurations

## Proposed scope
1. **New operation type:** `streaming-expression` in workload definitions, analogous to
the existing `query` and `bulk-index` operations
2. **Dedicated driver:** Execute expressions via `/stream` endpoint, collect timing and
tuple-count metrics
3. **Telemetry:** Capture rows-returned, execution-time, and error rate per expression
4. **Sample workload:** Add an example workload in `benchmarks/` that exercises common
streaming patterns (search, facet, rollup, jdbc-like aggregation)

## References
- https://solr.apache.org/guide/solr/latest/query-guide/streaming-expressions.html