I see no problem with SPV mining while verifying the previous block. It would also make my above suggestion pointless.
Yes.
Moreover, I don't see why the time spent verifying would be significant. The vast majority of transactions would have been verified by a full node before being included in the block, and then it is just a case of checking the hash, and looking up that the txn was previously mempool verified. The verification time for a block should be insignificant relative to its download time (unless it's a flood attack of new txns by a miner). Or am I missing something?
This was also implicit in my upthread point that the only justification for widespread 0 txn blocks would be bandwidth (propagation) delay.
IBLT purports to improve that by sending less impulse data when announcing the block, but it can't help if the steady state average rate of txns is greater than the bandwidth of the node and it can't stop the collusion of withholding transactions and flooding to force lower bandwidth nodes to be essentially SPV miners. That is why I say it is obfuscation of centralization.