feat: support configurable NQ for concurrent Milvus searches

[yanbinyang]

Fixes #873. Concurrent searches currently send one query vector per request and do not expose a per-request query batch size. VectorDBBench therefore cannot directly measure Milvus and Zilliz Cloud batch-search workloads with NQ greater than 1.

I added nq to the common case configuration and CLI (--nq, or nq in YAML), with the existing behavior preserved by a default of 1. Milvus advertises batch-search support through a client capability and implements one SDK search call for each fixed-size query batch; Zilliz Cloud inherits this implementation. The concurrent runner wraps through the query dataset to keep batches full, counts successful query vectors in qps and conc_qps_list, records latency per successful batch request, and excludes failed or incomplete batches from successful query-vector counts. Serial latency and recall remain single-query measurements, while unsupported backends and workloads reject nq > 1 before data loading.

A CLI unit test verifies that the configured NQ reaches the generated task configuration. Live Milvus validation used a custom runner's run_by_dur() path with NQ=64 on Cohere and BioASQ at 1M/10M and topK=10/100; all eight combinations reported no failed batches. This was not an end-to-end run through the standard CLI, and Zilliz Cloud was not live-tested.

[sre-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: yanbinyang
To complete the pull request process, please assign xuanyang-cn after the PR has been reviewed.
You can assign the PR to them by writing /assign @xuanyang-cn in a comment when ready.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[yanbinyang]

/assign @XuanYang-cn

Hi, when you have time, could you please take a look? The workflow is awaiting maintainer approval. Thanks!

[yanbinyang]

Hi @jamesgao-jpg, I have rebased this PR on main and added regression coverage in 40df7e3. All CI checks pass. When convenient, could you please take a look? Thanks!

[jamesgao-jpg]

/lgtm