This adds UnifiedDiffWriter.write_index(), which writes a standard

Index: ... line, followed by 67 = characters.
This is a fairly standard convention used by Unified Diff variants to

help separate files when there may not otherwise be a clear boundary.

For instance, when there may be metadata changes or binary file

differences, but no ---/+++ headers.
Like other methods, this accepts byte strings or Unicode strings,

handling encoding automatically.

Unit tests passed.
Used this in some in-progress work.