docs: collect false positive examples for human prose

[devswha]

Goal

Add a small false-positive gallery documenting human-written passages that patina may flag.

Why

patina should be positioned as an editing signal, not a truth machine. Documenting false positives makes the tool more credible and helps calibrate future scoring changes.

Suggested scope

• Create or extend a docs page with 3-5 false positive examples.
• Include the likely reason each passage is flagged: encyclopedic register, corporate tone, academic phrasing, etc.
• Mention that the score should be interpreted as an editing hint.

Acceptance criteria

• The examples do not accuse authors or identify private people.
• Each example explains why it is a false positive risk.
• README or FAQ links to the new section if a new page is created.