In the wine data, the very best and very worst scores have only a handful of wines, so their averages are shaky. A careful analyst only reports a score's average when enough wines got that score.
Task: write alcohol_by_score(scores, alcohol, min_wines). Wine i got quality score scores[i] and has alcohol level alcohol[i].
Return a dictionary mapping each score to the average alcohol of the wines with that score, rounded to 2 decimal places. Leave out any score that fewer than min_wines wines got.
{}.