Table download breaks for large number of rows

[pjcross]

To generate the other datasets for our scaling study, I'm trying to use a combination of filter choices to subselect rows that preserve our latin hypercube sampling. But when I try to download the visible rows, it fails. The row numbers are all being loaded into the URL, which then becomes too long. This needs to be bundled up into a post instead.