Planet spin-axis via RM effect?

The thought just hit me...

Assuming a sufficiently precise telescope, and a sufficiently quiet star, couldn't the Rossiter-McLaughlin effect be used in reverse? Shouldn't we be able to detect the alignment between the planet's spin axis and its orbit? I'm thinking something like a transiting hot Neptune orbiting a quiet M dwarf or something. I would expect that with the planet in a synchronous orbit, and thus rotating slowly, there probably wouldn't be a large signal (not to mention the stellar RV more than dominating the curve anyway).

In the secondary transit, there should be a dip up or down, depending on the spin axis (and assuming it is perpendicular to the orbit plane) between first contact and second contact, and then a mirror of that dip between third and fourth contact.

Any thoughts? (other than, of course, "good luck detecting that.")
It would be interesting to take an RV curve by those astro-combs that are supposed to bring about cm-accuracy or so, and subtract out the stellar RV, leaving only the RV of the planet.