In this case, where they know the previous block is valid b/c they mined it, why wouldn't they simply mine the next with tx's in it? the reason these spv miners do what they do is b/c they don't want to waste time validating another pools just received block, which is not the case in this situation since it's their own.
It is both as can be seen when 2 blocks back to back come from an SPV pool and the 2nd block has zero txns. The reason is they've not gone to the effort to make the bitcoind validation process fast in their own nodes and they don't want to wait for bitcoind to validate the block (it takes time) before starting on their next block. For solo.ckpool.org and kano.is we run a customised bitcoind which speeds up the validation process dramatically, making this delay negligible.
but again, in the Antpool example given of 2 of their own blocks found sequentially with the 2nd being an spv 0-1tx block, they already know the 1st block is valid. afterall, they produced it. why bother then with spv mining the 2nd block instead of just stuffing the 2nd block straightaway with tx's so as to capture fees too?
also, the root emphasis as to why they spv mine is to avoid constructing the getblocktemplate during stress conditions of full or nearly full blocks and bloated mempools:
https://www.youtube.com/watch?feature=player_detailpage&v=H-ErmmDQRFs#t=4330and here:
https://bitco.in/forum/threads/gold-collapsing-bitcoin-up.16/page-155#post-5252