It is not possible for an unmodified copy of web-decker to programmatically fetch external files hosted e.g. somewhere on github pages, make http requests, etc. Doing this sort of thing would require adding custom javascript extensions to web-decker. An example of how such an extension might be written can be found in this subthread.
In general, decks should be self-contained objects pre-packed with any data they require to operate. A deck is a database, with grid widgets for storing tables and Lil for querying and manipulating them.
Lilt can be used to import or export data to or from a .deck or .html file in an automated fashion. If you insist on storing some dataset independently from a deck that uses it, it might be possible to use Lilt as part of a continuous integration pipeline or "build script" to bake out an updated web-decker build when an associated csv file is changed.