The only way I've figure this out is to instead of name an artifact, put all of the artifacts into a "directory" as part of the build process, say "artifacts/", and define the artifacts as "artifacts/**". Then on the Deployment side, be clever about manipulating the artifacts for deployments.
Note, in my case, I have an issue with multiple branches for the same build (think, "future release", "current release", "legacy release") that may have different artifacts on them (either new features in "future release", or aged off artifacts from "legacy release"). I had to wrap the actual deployments into a script that was "smart enough" to just iterate through artifacts that actually existed for a given deployment environment.
I'm not completely happy with Bamboo's treatment of special cases for artifact management at all. In fact, I've found that judicious use of the "script" task in Bamboo (and managing those scripts in some external git repo) seems to be the only real way to manage larger Bamboo installations in general.
