[gjm11]

I've only glanced at the paper, but it looks to me as if he's using it in another perfectly usual way, namely referring to situations where the wavefunction and/or the Hilbert space it lives in can be written exactly or approximately as a product over simpler things. There's nothing wrong with that.