On this issue of combining accents or single pre-combined glyphs, Andrew Ollett has already stated the issues. The Unicode Consortium's documentation about this business is at
As Andrew said, the terminology is "Normalization Form," and there are several of these, and quite a few points to consider when writing programs to work with Unicode. The Unicode consortium makes available algorithms for dealing with all this, and modern text-processing software libraries for Unicode are commonly aware of Normalization Forms and "do the right thing," so the end-user doesn't have to worry.
This doesn't always work, though. I find that when I cut-n-paste from WorldCat, for example, into JabRef, all the accented letters are retained in NFD format, and it's annoying.
Mostly, when we talk about typing IAST with pre-composed, single-glyph characters like ā, we're doing what the Unicode people call "Normal Form Composed" or NFC. The files in SARIT and elsewhere are in this format.
Best,
Dominik