There is actually an open ticket (#12) for this. In theory I'd like to integrate something like this, but that would require some serious thoughts about how a good design for this would look like. It would have to be efficient (performance/memory) and backwards compatible, and ideally it would even reduce the amount of memory required per request.
So currently there are no concrete plans, because the string only solution is sufficient in most cases and the design isn't clear yet.
maybe something like this:
res.headers is still a DictionaryList.
res.headers.std.headerName is used for accessing standard headers.
e.g. res.headers.std.contentDisposition.filename would return a slice (possibly empty) providing access to some of the chars in res.headers["Content-Disposition"]. obviously header bodies should be parsed in a lazy manner.