Shrink searcher thread pool

[rdettai-sk]

Description

Reduce the searcher thread pool size by 1 by default. We have seen cases where tasks are storming the thread pool queue and the whole searcher becomes irresponsive. In deployments on big machines, leaving one CPU free is acceptable and gives the tokio runtime more room to keep things running smoothly.

Also add caller to searcher thread pool metrics.

[Copilot]

Pull request overview

This PR adjusts how CPU-intensive search work is scheduled and observed by (1) shrinking the default searcher Rayon thread pool size by one CPU and (2) adding a caller label to thread pool task metrics, so CPU task sources can be distinguished in observability.

Changes:

• Configure the search thread pool to default to num_cpus - 1 threads (with an env override to use Rayon’s default sizing).
• Add a “caller” identifier to thread pool CPU task metrics and update search code paths to pass caller IDs.
• Refactor some CPU-task closures for readability while preserving behavior.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
quickwit/quickwit-search/src/root.rs	Adds identified caller when merging leaf search results on the search thread pool.
quickwit/quickwit-search/src/list_fields.rs	Adds identified caller labels to list-fields merge CPU tasks and refactors the CPU closure.
quickwit/quickwit-search/src/lib.rs	Changes search thread pool sizing defaults and logs configuration at init time.
quickwit/quickwit-search/src/leaf.rs	Adds identified caller labels for several CPU-intensive leaf search tasks.
quickwit/quickwit-common/src/thread_pool.rs	Introduces per-caller metrics labeling and adds a new API for identified CPU tasks.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[rdettai-sk]

lot of diff for nothing (just changed run_cpu_intensive to run_cpu_intensive_with_identified_caller)