I’ve been looking through the proof of concept code for Halo. It’s very promising, but far from production ready now.
For reference, Zcash has a network upgrade planned for December 11th of this year that doesn’t include any improvements to the zkp scheme used in Sapling. After that we should see their timeline for the next upgrade. There are benefits to tracking them, evidenced by the counterfeiting vulnerability they patched last year.
It’s also worth noting that Halo isn’t a theoretical performance improvement over the scheme used in Sapling when it comes to shielded transactions. It’s designed for efficient recursive proofs such as those necessary for scaling, as explored in Coda and Rollup. So as exciting as it is to be able to drop the trusted setup, perhaps best to think of Halo as useful for different applications.