Coverity complains about an index being unset (CID 1356740). While I'm
pretty sure this is a false positive given the code path in question
there's no performance reason not to set the index here.

While I was at it, I also fixed a GCC warning about a separate
"potentially uninitialized" variable. Again seems to be a false positive
(unless an extract operation starts somehow already marked to abort) but
fixing anyways. I chose ARCHIVE_RETRY to hint to *not* abort the
operation if this does somehow make it into error path.