auto-calculate --users from --requests_per_sec to target QPS is achieved

[zeke-emerson]

Problem

When --requests_per_sec is set but --users is left at its default of 1, the benchmark often can't achieve the target QPS. Each user issues requests sequentially, so a single user's max throughput is bounded by request latency (1 / latency). Users have to manually calculate and set --users to work around this, which is error-prone.

Solution

Auto-calculate --users from --requests_per_sec when not explicitly provided. Assuming a conservative 500ms request latency, each user can do ~2 req/s, so we provision ceil(qps * 0.5) users. Explicit --users still overrides the calculation. When --requests_per_sec is not set (unlimited), defaults to 1 user as before.

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update
• Infrastructure change (CI configs, etc)
• Non-code change (docs, etc)
• None of the above: (explain here)

Test Plan

• Run with --requests_per_sec=100 and no --users: startup log shows users=50
• Run with no --requests_per_sec and no --users: startup log shows users=1 (backward compat)
• Run with --requests_per_sec=100 --users=10: startup log shows users=10 (explicit override)
• Run with --requests_per_sec=1 and no --users: startup log shows users=1 (low QPS edge case)

[ajit-vijayakumar]

LGTM