support extended statistics for column groups #18330

Development Task

To avoid independent assumptions of multiple columns in planner estimation, provide interfaces to collect extended statistics as required, e.g:

1) CREATE STATISTICS stats1 (cardinality) ON t(a,b);
2) CREATE STATISTICS stats2 (dependency) ON t(a,b);
3) CREATE STATISTICS stats3 (correlation) ON t(a,b);

1 helps for cases like select * from t group by a, b / select * from t join tt on t.a = tt.a and t.b = tt.b;
2 helps for cases like select * from t where a = 1 and b > 1;
3 helps for cases like select * from t where a > 1 order by b limit 1;