estimation is zero when there is no buckets #59563
Description
Enhancement
from the code here, we can see that when there is no histogram, the estimation should go to branch1, while from the stat loading from below,
column.hist.NDV=31, the column.topN.num()=28, and there is no buckets at all. This is mainly because hist.NDV is calculated from all rows before sampling. The case exists when sampled rows are all in the topN, while the total NDV is greater than sampled rows.
