Clustering side-effects

[Robert Leach]

I screen-recorded a long debugging session. These side-effect issues of clustering are visible at around time index 3:34. If you need to see it, you can download the 500MB file here:

http://gen-rlimac.princeton.edu/~rleach/TREEVIEWBUGS/multiple_bugs/multiple_bugs.mov

These issues may have to do with being zoomed in while clustering:

1. The rows and columns are re-ordered, as you would expect. The row/column labels change. The data changes. The previously blue/selected column headers disappear and are replaced with black unselected column headers - this is all well & good, but the yellow selection area remains. The selection should probably disappear as well.

2. The row label type changes from "COMPLEX" to "GID". Clustering should not change the label type of the rows.

[Christopher Keil]

Everything is deselected upon opening the clustering window. Selected label type names are stored before a new file is loaded so that their index can be found in the newly loaded file's HeaderSummary. The list of names in HeaderSummary changes when clustering, so the indices shift. This is now accounted for.