This is a reasonable model, and what is true of HTLV-1 is likely to be more true for HTLV-2, which appears better adapted to human hosts.
This also has a bearing on failures to culture. Unless you get lucky, you likely need dividing cells for viral replication. This was not in the original Science paper, and is an easy target for critics. (Check on the value of royalties for HIV test kits, and patent filings concerning XMRV tests, if you wonder about driving forces behind disputes over tests. Priority is of more than academic interest.)
One caution, slow mutation after insertion says nothing about mutation of RNA prior to transcription. I'm still betting on considerable variation from this cause. I will also go out on a limb and say some endogenous sequences may be recent. Molecular clock arguments don't consider mutation prior to insertion. Arguments about descent from a common ancestor would need a better understanding of the population distribution of insertion sites than we have at present. Horizontal transmission followed by insertion of similar sequences could look like descent from a common ancestor, if you don't know where it was inserted.
We need better ways of recognizing virus without simply excluding anything that resembles an endogenous sequence.