So the way (I think) it works is that Tesla sets up a limited liability company that is fully owned, operated and consolidated by Tesla. The leases in question, both the financial instruments (ie, income from the customers) and the assets (ownership of the car) are transferred to this company. The subsidiary pays Tesla some $500M for this. To pay the $500M the subsidiary writes a bond, entitling bond holders to all future income from this particular company (and also obviously clauses to make it operate its business in a way that is in direct interest of the bondholders). So while Tesla has sold the value of the remaining depreciated car at the end of the lease, the bond holders will not own that car, they just have a payment due from the limited company that now owns the car.
If at some point, residual values are higher and/or defaults lower than initially estimated, the limited company will make a profit. I believe this profit will revert to Tesla. If on the other hand there is a shortfall then that limited company will not be able to pay bond holders their final tranches. But Tesla will just walk away. So that''s why, even if the assets are fully controlled by Tesla, it is still the bond holders that have the most interest in making sure that their residual values are estimated correctly.
I suppose since this is Tesla's first round, bond holders are looking for a lot of safe side. Ie, they'll be demanding that residual values are estimed lowly and defaults highly. In addition to a fat interest rate. Possibly this makes it more expensive for Tesla to sell out this particular batch of cars than to just keep it on the books. But I believe the real value to Tesla is to get started with it so future potential bond buyers understand the economics of leased Tesla cars, are able to predict residuals and default rates better and over time bid down between them the safety margin they ask from Tesla.