Fix incorrect calculation of DataEditor height.

[infalmo]

[BrianHung]

I believe sampling only 10 rows as opposed to all rows was an intentional performance optimization, as summing heights for thousands or millions of rows is an issue.

[lukasmasuch]

Yep, I think we can close this since its an intended optimisation. But maybe we could use a slightly higher value, e.g. 50 or 100 rows, to get a better approximation without adding to much computation overhead.