fix(core): avoid overall quantifier false positives

[coyaSONG]

AI disclosure: This pull request was authored autonomously by OpenAI Codex. I reviewed Harper's contributing and agent policies, issue #1249, the related closed PR, the implementation diff, and the validation output before submission.

Issues

Fixes #1249

Description

The Overall closed-compound rule previously suggested overall for every occurrence of over all. In normal phrases such as “iterate over all the save data,” however, over is a preposition and all is a quantifier.

This change gives Overall a small context-aware wrapper and emits the existing MapPhraseLinter correction only when a preceding determiner establishes the adjectival use, as in the existing “The over all performance...” positive test. It adds the two issue examples plus quantifier cases covering a plural noun, a pronoun, and sentence-initial usage. Regenerated corpus snapshots remove four existing false positives from the U.S. Constitution, Alice's Adventures in Wonderland, and The Great Gatsby.

Related: #3455 was an earlier attempt that special-cased words following over all; its author closed it. This patch uses grammatical context and covers the additional false positives already present in Harper's corpus snapshots.

Demo

Not visual.

How Has This Been Tested?

• just format
• cargo test -p harper-core -- closed_compounds::tests (40 passed)
• cargo test -p harper-core --test linters -- test_most_lints (snapshot regeneration followed by a clean pass)
• cargo test -p harper-core --lib (5,526 passed, 284 existing ignored, 0 failed)
• cargo clippy -p harper-core --lib -- -D warnings -D clippy::dbg_macro -D clippy::needless_raw_string_hashes
• cargo check -p harper-core
• just test-rust (full Rust workspace, 0 failed)
• git diff --check

AI Disclosure

• I am a human and didn't use any AI.
• I used LLM features of my editor, but not an agent.
• I used an AI agent interactively.
• I am an agent or I got an agent to do the work autonomously.

If Your PR Implements or Enhances a Linter

• I made up the sentences in the unit tests.
• The sentences in the unit tests were generated by an AI.
• I'm using examples from the bug report / feature request.
• I collected real-world sentences for the unit tests.

Checklist

• I have performed a self-review of my own code
• I have added tests to cover my changes
• I have considered splitting this into smaller pull requests.

[hippietrail]

I've only had a cursory glance at this and the results look good but I haven't analysed the logic yet. I would like to be explicit about the name of one test needing "of", which is very very minor.

[hippietrail]

I would call this allow_over_all_of_pronoun.