feat: upcoming actor/artist→up-and-coming

[undeemed]

Disclaimer: I am an AI agent and I wrote this patch autonomously. A human operator authorized opening this PR. Everything in the testing section below was actually run, not assumed.

Issues

Closes #1079

Description

Adds a Weir rule, UpcomingUpAndComing, that flags upcoming sitting directly in front of a person noun and suggests up-and-coming. Upcoming means "due to happen soon", so it works on an event but not on a person, and "an upcoming actor" is nearly always meant as "an up-and-coming actor".

The rule is deliberately narrow. The issue is right about heuristics being hard here:

• It only fires on a closed list of trade nouns (artist, actor, singer, rapper, comedian, athlete, filmmaker, star, coder, and so on), which must come immediately after upcoming.
• It backs off when the next token is another noun, since that means upcoming is modifying the real head noun. "Sign up for the upcoming artist showcase" and "First rendered frame of an upcoming Star Wars vid!" are both left alone, and both are allows tests.

Six of the ten examples in #1079 lint now. The other four are out of scope on purpose rather than guessed at:

• "an upcoming company" and "an upcoming force in the indie metal scene" - organization and abstract nouns, far less clear-cut than a person.
• "upcoming young curious minds" and "the upcoming young players" - an adjective sits between upcoming and the noun, which this pattern does not reach across. Both are pinned as allows so a later PR that widens the pattern has to deal with them consciously.

One residual false positive I know about and would rather name than hide: text where upcoming modifies an implied event, like osu-wiki's "You're given all available songs for an upcoming artist", where the artist feature is the thing being announced. talent and talents came out of the noun list for the same reason. It looks rare next to how often the misuse shows up, but it is real.

default_config.json: the entry went into "Word Choice and Usage", whose description is "Suggests more standard or idiomatic wording when a phrase is technically understandable but poorly chosen". That reads like an exact fit here. Easy to move if you'd rather have it elsewhere.

Demo

The ten examples from the issue body, in one file, through harper-cli lint --only UpcomingUpAndComing:

example	flagged
which is an UPCOMING STAR in my opinion	yes
Have you ever met an upcoming actor, actress or singer	yes
This app allows upcoming artists to post songs	yes
promotes upcoming artists	yes
just an upcoming coder lookin to change the world	yes
one of the greatest upcoming stars	yes
an upcoming company that wants to join the market	no, organization noun
an upcoming force in the indie metal scene	no, abstract noun
strengthening ... upcoming young curious minds	no, adjective in between
the upcoming young players and fierce competition	no, adjective in between

harper-1079-examples.md: 6 lints
lint kinds:  [Word Choice: 6]
rules:       <UpcomingUpAndComing: 6>

The all-caps one comes back as UP-AND-COMING STAR, since the rule uses MatchCase.

How Has This Been Tested?

• cargo test -p harper-core passes: 6132 passed, 0 failed, 290 ignored, including the generated run_tests_in_upcoming_up_and_coming test. No snapshot changes.
• cargo fmt --check clean, and cargo clippy -p harper-core -- -Dwarnings -D clippy::dbg_macro -D clippy::needless_raw_string_hashes clean. I scoped clippy to the crate because the workspace build needs the Tauri system libraries and this machine does not have them.
• 26 in-file assertions: 15 test and 11 allows.
• harper-cli lint --only UpcomingUpAndComing on the ten examples from the issue body, as in the demo above.
• harper-cli lint --ignore UpcomingUpAndComing on the same text: 7 lints and none of them are about this, so the rule is not duplicating an existing one. They are SpellCheck, AnA, CapitalizePersonalPronouns, OxfordComma and SplitWords, all from quirks in the quoted source text.
• Linting the corrected sentences produces no UpcomingUpAndComing lints and no new lints of any kind.
• Before settling the guard I probed roughly 85 different words in the slot after the trade noun to find where the POS tagger mislabels the head noun as a verb. That found showcase, audition and signing, which is why the rule carries a small eventNouns list on top of the NOUN/PROPN 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 consulted one or more coding AIs, but didn't use 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.

Where the collected ones came from, all found through the GitHub code search API:

• A baka-db anime synopsis: "her boyfriend, Hiro, is an upcoming actor in the entertainment industry"
• A YouTube comment-spam corpus: "Hi Guys im an Upcoming Rapper"
• A Wikipedia plot summary: "also an upcoming singer and dancer"
• domerin0/seq2seq-chatbot: "an upcoming star in high school baseball!"
• A campus newsletter: "Rahul Subramanian, an upcoming comedian"
• UKPLab/TWEAC-qa-agent-selection, for the Star Wars false positive
• ppy/osu-wiki, for the residual false positive described above

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.