If jday were guaranteed to be non-negative, the first could also be changed to

  jday = std::round(jday);

but that would give different results for negative jday.  So play it safe by
assuming that jday might be negative, and that if it were, the original way of
rounding was actually intended.

Change-Id: I137ee659e738791a7d260ce86cba60cdbbc084a6
Reviewed-on: https://gerrit.libreoffice.org/48004Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>