Add likert_question type: survey matrix of radio buttons

[Copilot]

Adds a new likert_question type — a matrix where multiple statements share a common scale, each answered via a radio button row.

Data format

Instructions:

{
  "type": "likert_question",
  "statements": ["I enjoy Python.", "Tests are useful."],
  "options": ["Strongly Disagree", "Disagree", "Neutral", "Agree", "Strongly Agree"],
  "points": 2
}

Checks (optional — omit correct entirely for survey/always-correct mode):

{
  "correct": {"0": "Agree", "1": "Strongly Agree"},
  "wrong_any": "Some selections were unexpected."
}

Student answer: {"0": "Agree", "1": "Neutral"} (string-keyed dict, one entry per statement)

Changes

• questions.ts — adds likert_question to QuizQuestionTypes; student state is a {[index]: ko.Observable<string>} dict (same pattern as multiple_dropdowns_question)
• quizzes.py — grading branch: partial credit (correct / total); no correct key = survey mode, always full credit
• questions_ui.html — <table> with option column headers, one radio-button row per statement; binds checked to student[String(statementIndex)]
• quiz_editor_state.ts — adds likert_options, likert_correct, likert_wrong_any fields; toInstructionsJson/toChecksJson serialise correctly; addStatement/removeStatement keep likert_correct in sync
• quiz_editor_ui.html — editor form: statement list, scale options list, per-statement correct-answer dropdown (leave blank = survey mode)
• tests/test_quiz_grading.py — 8 new tests covering all-correct, all-wrong, partial credit, wrong_any feedback, survey mode, empty/partial student answers
• quiz_schema.md — documents format, grading rules, and survey mode