pre-aggregation rework

[shawnlaffan]

Pre-aggregation of data is a useful thing (see #16) but is likely best applied to existing data frames, rasters and similar objects.

We can let the Biodiverse itself handle files on-disk.

While we are at it, the current code needs to:

• [ ] Handle multiple label columns
• [ ] Use the same quoting rules as Biodiverse for labels and groups
• [ ] Handle the sample count columns