Implement a meeting scheduler that stores meetings as half-open intervals `[start, end)`, where `start` and `end` are integer timestamps and `start < end`. The scheduler must support: 1. `book(start, end) -> bool`: Add the meeting if it does not overlap with any existing meeting. Return `true` if the meeting is booked, otherwise return `false` and leave the schedule unchanged. 2. `remove_before(time) -> int`: Given a cutoff timestamp, remove all meetings that end at or before `time`. Return the number of removed meetings. 3. `get_meetings() -> List[Tuple[int, int]]`: Return all currently scheduled meetings sorted by start time. Follow-up discussion: - What data structure would you use to optimize `book` and `remove_before`? - How would the complexity change if you used an unsorted list, a sorted array, or a balanced binary search tree? - What edge cases should be covered by tests? Example: ```text book(10, 20) -> true book(20, 30) -> true # Adjacent meetings are allowed. book(15, 25) -> false # Overlaps with [10, 20). book(5, 10) -> true remove_before(20) -> 2 # Removes [5, 10) and [10, 20). get_meetings() -> [(20, 30)] ``` Assume the number of meetings can be large, so the solution should be efficient.