To elaborate a bit, it looks like filenames supplied for -bucket and -cbucket get misparsed when they include suffix, extension, additional .gz extension, or some combination thereof.
These are the files created:
Decon+orig.BRIK.gz+orig.BRIK (27M--not compressed)
Decon+orig.BRIK.gz+orig.HEAD (200K)
Decon+orig.BRIK.gz.xsave (2.2M)
Decon+orig.xmat.1D (596K)
Decon_coefs+orig.BRIK.gz+orig.BRIK (18M--not compressed)
Decon_coefs+orig.BRIK.gz+orig.HEAD (11K)
Decon_res+orig.nii.gz (126M--compressed)
For whatever reason it actually parses -bucket correctly for the purpose of naming the design matrix 1D, but not the other outputs based on it. (I'm not sure why it would be parsed more than once, but....)