Unfortunately your screenshots render low resolution though they are 1680×1050 which makes very difficult to see details.
Index entries are sorted relative to Unicode encoding. This means precomposed letters with diacritics usually appears after z. It is the case for U+0101 LATIN SMALL LETTER A WITH MACRON. There is another representation as U+0061 LATIN SMALL LETTER A followed by U+0304 COMBINING MACRON which would then sort between “a” and “b”.
However this dual representation creates problems and the Unicode Consortium has defined a text normalization procedure to deal with them, producing either the precomposed form or th fully decomposed form.
That said, I don’t know what Writer does internally. Solving for your requirement of sorting words on their base character calls for fully decomposed form. Then you should replace the precomposed glyphs in the concordance file by their canonically equivalent decomposed form as base letter followed by combining diacritical mark.
It is possible that you may need to do the same in your text (fortunately, this can be done with Find & Replace).
I have checked that the fully decomposed form sorts as you wish, but see my caveat about how words are typed in the text (precomposed vs. fully decomposed).