I guess epidemiology has coopted the term. In the study of evolution it means separation of a gene pool from another, usually geographically, and explains the different racial groups on the planet, along with other three components of evolution (mutation, mixture, and natural selection).
Applied to ebola, it would explain different strains appearing in different countries, for instance, or an airborne strain developing in one area.
As for viruses mutating within a single person, yes, that is a big danger, because of the heavy viral load of this disease, but is only tangentially related to genetic drift as it has usually been understood.
The definition of genetic drift is the same, it is just that the time scale of the process in viruses is very short compared to the time scale when discussing large organisms. Also, the viruses don’t trade genetic material with each other, which further accelerates the process of genetic drift, since frequency of alleles is not a measure of drift.
The recent Science paper with the sequencing data shows phylogenetic trees; one of the trees (figure 3) is rooted back in February. It shows a single source for the outbreak, but the virus has already separated into two distinct clusters in Sierra Leone and Guinea, and each of those clusters appears to have sub clusters.
http://www.sciencemag.org/content/early/2014/08/27/science.1259657.full