PromQL support in ES|QL (GA)

## Value Prop

PromQL is a widely used query language for metrics and time-series analysis, and many engineers already build their metric queries, dashboards, and alerts with it. Native PromQL support in Elasticsearch lets those users analyze their metrics with the language and skills they already have — directly within ES|QL and Kibana, guided by editor autocomplete and syntax support — instead of learning a new language or switching to a separate tool for metrics analyses.

The result is a single, unified experience: engineers can work with metrics in PromQL alongside their logs and traces on the same platform, correlate across all of their observability data in one place, and cut out the friction and context-switching of moving between tools — leading to faster investigations and simpler operations.

## Expected Outcome

PromQL shipped in Tech Preview in Elasticsearch 9.4, where users can already write PromQL through the `PROMQL` command in ES|QL and use PromQL in Kibana.

For GA, we are taking PromQL support to production readiness and significantly expanding language coverage:

- **Broader language coverage:** substantially expanded support for PromQL functions and query patterns
- **Query endpoints:** support for instant, range, and metadata queries
- **Functions & histograms:** `histogram_quantile`, with support for both legacy and native (exponential) histograms
- **Correctness & consistency:** accurate, consistent results, including correct time bucketing and bucket labels
- **Metric handling:** refined metric classification to avoid false-positive query rejections (e.g. allowing `rate` on gauge metrics)
- **Stability:** production-ready quality with no high-severity correctness or stability issues, including support for downsampled and mixed data sources
- **Scale & performance:** scalability and performance validation for large metric workloads

Support we plan to add after GA:

- Vector matching and group modifiers (`on`, `group_left`, `group_right`; one-to-one and one-to-many)
- Additional functions such as `topk` / `bottomk`, `label_replace`, `label_join`, and `sort`
- Subqueries and binary set operators
- Additional query endpoints and APIs