Letter board: fix the failing check, then pick words that use the most distinct letters

Context

Welcome to the store design team. The shop window's letter board has one tile per letter, and letterboard/ is meant to plan which product names to spell on it; the rules are in README.md. Start by running the tests: the check for a pick fails, and that comes first. Then build the planner.

You have 40 minutes. The checks are graded one by one, so each part you get working counts even if the rest is not done. Use the AI assistant if you like, but check what it gives you against the README and the tests.

MU-1: The check accepts picks it should not

Report

tiles_used(["moon"]) says 3. The board has one o, so moon cannot be spelled at all.

Acceptance criteria

• tiles_used returns 0 for any pick where a letter appears twice, in one word or across words.

MU-2: The most tiles (level 2)

Acceptance criteria

• most_tiles(words) returns the most tiles any pick can use.
• A list of 30 words comes back well within a second.

MU-3: The pick (level 3)

Acceptance criteria

• best_pick(words) returns the words of a best pick in list order, with ties broken by list position as the README describes.

Notes

• Run pytest tests/visible. Three tests fail on the current code.
• The hidden checks include lists of 30 words, with a time limit.
• Practice task modelled on candidate reports of Meta's AI-enabled coding round, where a core problem was picking words to use the most unique characters. Original StackYard code, not the actual question; not affiliated with Meta.