docs: document scalingFactor timer resolution and memory growth pitfalls

[vinayjn]

Description

Add guidance under the scalingFactor section for two common issues:

• Timer granularity on Apple Silicon (~41.67 ns at 24 MHz) makes sub-100 ns benchmarks unreliable without an inner loop. Recommend .kilo for 1–500 ns operations and .mega for sub-nanosecond work.

• Stateful mutation benchmarks with high scalingFactor can exhaust memory (e.g. .kilo × 10K iterations = 10M mutations). Show two mitigation patterns: capping maxIterations and using startMeasurement() to exclude periodic resets.

How Has This Been Tested?

Documentation-only change. Verified DocC renders correctly locally.

Minimal checklist:

• I have performed a self-review of my own code
• I have added DocC code-level documentation for any public interfaces exported by the package
• I have added unit and/or integration tests that prove my fix is effective or that my feature works

[gemini-code-assist]

Code Review

This pull request updates the documentation for writing benchmarks by adding sections on the scalingFactor and handling stateful benchmarks. The new content explains how to mitigate timer resolution issues on Apple Silicon and prevent memory exhaustion during high-iteration tests. Review feedback suggests clarifying that the benchmark closure is responsible for executing iterations, correcting the performance description of Date() creation, and highlighting that resetting state within the closure is the preferred method for maintaining measurement stability.