Minimize Manhattan Distances — C++

// Time:  O(n)
// Space: O(1)

// math
class Solution {
public:
    int minimumDistance(vector<vector<int>>& points) {
        const auto& max_distance = [&](int exclude) {
            static const int POS_INF = numeric_limits<int>::max();
            static const int NEG_INF = numeric_limits<int>::min();
    
            pair<int, int> max_sum = {NEG_INF, -1};
            pair<int, int> min_sum = {POS_INF, -1};
            pair<int, int> max_diff = {NEG_INF, -1};
            pair<int, int> min_diff = {POS_INF, -1};
            for (int i = 0; i < size(points); ++i) {
                if (i == exclude) {
                    continue;
                }
                const int x = points[i][0], y = points[i][1];
                max_sum = max(max_sum, {x + y, i});
                min_sum = min(min_sum, {x + y, i});
                max_diff = max(max_diff, {x - y, i});
                min_diff = min(min_diff, {x - y, i});
            }
            return max(tuple{max_sum.first - min_sum.first, max_sum.second, min_sum.second},
                       tuple{max_diff.first - min_diff.first, max_diff.second, min_diff.second});
        };
    
        const auto& [_, i, j] = max_distance(-1);
        return min(get<0>(max_distance(i)), get<0>(max_distance(j)));
    }
};