Add benchmark/integration fixtures for rules-nlp

[Peyton-Spencer]

Summary

Add benchmark/integration fixtures for rules-nlp so realistic marketing-copy samples can catch diagnostic drift across releases.

Proposal

Create a packages/rules-nlp/test/fixtures/ directory with representative copy samples, then add an integration test that runs the rule suite over each fixture and asserts the diagnostic count stays stable.

Fixture Coverage Ideas

• SaaS homepage hero copy
• Product launch announcement
• Enterprise feature page
• Lifecycle email or onboarding copy
• Short social/ad copy

Expected Test Shape

• Load fixture files from packages/rules-nlp/test/fixtures/.
• Run the configured rules-nlp registry against each fixture.
• Assert diagnostic counts per fixture and, where useful, per rule.
• Keep expected counts explicit so intentional rule behavior changes require deliberate snapshot/count updates.

Risk Note

Fixture tests can become noisy if they assert too much detail. Start with stable diagnostic counts and only add exact range assertions for high-value regression cases.

Implementation Reference

Use existing rules-nlp tests and no-buzzword-stacks coverage as the wiring reference:

https://github.com/omniaura/faircopy/blob/main/packages/rules-nlp/test/rules.test.mjs