Implement a Trie and support prefix search + delete

A trie node stores a map or fixed alphabet array of children plus isWord; optionally store a subtree word count for efficient prefix counts. Insert walks characters, creating missing nodes, then marks the final node. Exact search succeeds only when traversal ends at a terminal node, while prefix search succeeds once the prefix path exists and can enumerate descendants with DFS if actual completions are needed. Deletion first verifies the word exists, clears its terminal flag, then walks back upward removing a child only when that child has no children and is not terminal for another word. This bottom-up rule preserves shared prefixes such as car and cart. Clarify duplicate insertion semantics, case folding, Unicode handling, and whether prefix search returns a boolean, count, or list. Test deleting a word that is a prefix of another, deleting a missing word, empty string policy, and repeated insertions. Operations are O(L) for word length, plus output size for enumeration; storage is O(total stored characters).