Still thinking about ways to link to supplementary data from papers and store that data locally.
A long time ago Microsoft used to indicate the location of a CDF XML file (a precursor of RSS) which would itself contain links to files that should be downloaded and cached by Internet Explorer when a page is made available offline. The modern equivalent is rel="enclosure", which indicates—essentially—links to supplementary data. rel="enclosure" can be used in Atom or RSS (particularly in podcasts) and (X)HTML.
MAF is a Firefox extension that "allows complete web pages to be saved in a single archive file. MAF stands for Mozilla Archive Format and the extension uses RDF to save page meta-data such as the original URL of the page and the date/time the page was put in the archive." If MAF could be made to support rel="enclosure", so it would download linked files and save them all in the same archive, that would be ideal. The only problem then would be finding tools that could read the archives (though they're just zip archives, so that shouldn't be too difficult).