Apologies for the delayed follow up, been a bit of a fire fighting exercise as of late.
When I say the same, I mean 100% the same. I pulled down a copy of the database locally to play with both. Locally, it took to the new index immediately, Azure ... did not change, even after updating stats.
Oddly enough, after we did a DBCC STACKDUMP (to move hosts), it all of the sudden liked the new index and began to use it! This also cleared up overall performance issues we were seeing (this view was taking 12+ hours and timing out, now it takes ~15 minutes).
Yet now this morning, after we had another failure ... it's back to using the old index, and the performance is back to being a turd (with no other contention and very minimal utilization). Still playing, but definitely some oddities here.