My impression (which I need to confirm by looking at the actual blockchain) is that other than standard multisig scripts, there is negligible use of nontrivial scripts on the Zcash network. So, if we support shielded multisig then that’s sufficient to unblock the removal of t-addresses.
There are proposals, such as P2VK and Zexe, for supporting arbitrary proof statements in a Zcash-like cryptocurrency. This is in principle more flexible than the Bitcoin script system and could support Ethereum-style smart contracts, where (unlike Ethereum or Bitcoin) the statement being proven could be completely private. However, in my opinion just dumping support for that into the low-level consensus protocol without the necessary tooling to write secure contracts would be at best unusable, and at worst a recipe for encouraging people to lock up their money into contracts for which any tiny bug would cause it to be lost. I would expect a few more years of improvement of the tools around zk proof systems to be necessary before Zcash adds something like that.
Note that support for user-issued tokens, which seems to be the main thing people use Ethereum smart contracts for, does not actually require smart contracts.