1.  when an individual hasher joins a pool large or small, he will not make earn any more BTC or reduce any orphans he might have gotten otherwise while mining individually, he just smooths both of those out over time.  this assumes that both large and small pools are joining the relay network (it's open so why not) or otherwise have excellent internet connections which i have heard many miners claim.
2.  "Verifying the new tx incoming blocks contain has nothing to do with tx in mempools, which are waiting to be included in subsequent blocks."  such a statement is ignorant.  all the tx's received in the block results in the mempool being cleared of those same unconf tx's.
3.  your argument completely misses the point as well as the reality.  the answer is that it is exceedingly unlikely that Satoshi perfectly chose the point at which the network in July 2015 would consider 1MB to be "too large" to the pt that pools would start SPV creating hard forks.  that is so ridiculous on it's face it's not worth arguing.  also, it also totally ignores the fact that the top 5 Chinese pools have already gone on record stating that despite their problems with the GFC, they are perfectly ready and willing to inc the cap to 8MB and double it every 2 yr according to Gavin's proposal.  which says they are doing SPV mining for a reason other than the blocks are "large".  i can't say for sure b/c afaik, they haven't said exactly why they are SPV'ing.  i think it's b/c the cap is being continuously hit resulting in bloating unconf tx sets which is somehow affecting their willingness to include confirmed tx's in blocks.