Improve ImageJ's handling of large (>4GB) TIFF files

Nobody has claimed this yet.

It should be possible to quickly load, browse and save very large TIFF files in a coherent way. There are several pieces to this:
- [ ] Ensure SCIFIO transparently writes to BigTIFF when exporting using File > Export > Image...
- [ ] Make SCIFIO able to import ImageJ1's "raw pseudo TIFF" format
- [ ] Make SCIFIO's "virtual stack" (cell image) performance as good as possible
- [ ] Make SCIFIO recognize signs (e.g., an ImageJ1-style ImageDescription in the first IFD) when a TIFF is written monotonically, so it can be parsed more quickly (i.e., without explicitly reading all the IFDs)
- [ ] Make SCIFIO add such signs (probably some new metadata as a directory entry in the first IFD) when writing TIFFs

Also critical is to improve performance of Bio-Formats `setId` calls of various high-impact proprietary file formats such as ND2, although that work is outside the scope of this issue.

See also these mailing list threads:
- [Problems with Saving Tiff greater than 4 GB & bigTiff exporter](http://imagej.1557.x6.nabble.com/Problems-with-Saving-Tiff-greater-than-4-GB-bigTiff-exporter-tp5010074.html)
- [Processing huge files](http://imagej.1557.x6.nabble.com/Processing-huge-files-tp5012001.html)
- [bfconverter nd2 to tif](http://imagej.1557.x6.nabble.com/bfconverter-nd2-to-tif-tp5012133.html)

Review the SCIFIO TIFF import/export and virtual-stack work described in the checklist, starting with File > Export > Image... and the raw pseudo TIFF and monotonic-TIFF cases. Break the issue into independently testable parts and identify focused performance checks for large files; completion requires addressing the checklist items, while Bio-Formats setId performance is explicitly out of scope.

Written by the indexing model from the issue text.