Zeno,
The existence of biblio-koha-indexdefs.xml is not related to GRS-1 right now. It might have been (for a while) a way to transform the record.abs file into something translatable to XSLT. But it is a more useful thing now.
The XML file provides with a user-friendly syntax for defining indexes (and now also facets). The etc/zebradb/xsl/koha-indexdefs-to-zebra.xsl file is designed to process that syntax for creating a suitable XSLT for use in DOM.
So, if you want to change some indexes, or facet definitions: do it on the XML file.
If you want to change the behaviour of DOM indexing (for example, you could want to generate an index for the 'availability' filter on each record, so it is calculated on indexing time instead of complex un-doable PQF queries), then you should patch the meta-XSLT (koha-indexdefs-to-zebra.xsl) so it generates an XSLT that generates that specific index out of a MARCXML record.
It's been a while since we stopped re-generating the DOM files from the GRS-1, as we've been forcing people to fix both GRS-1 *and* DOM each time.
Hope this clarifies the landscape.
Regards
Tomas