The other reason, and probably the biggest reason, is the lender's accountability for their use of the appraisal. The banking regs hold the lenders directly accountable not only for their use of the appraisals but also for their selection of the appraiser. As a practical matter the gov't can't hold the lenders accountable to this degree without giving the lenders the discretion to accept or reject. The further the lenders are able to distance themselves from these decisions the harder it becomes to hold them accountable for what they're actually doing.
Moreover, the only intended use of these appraisals is to make mortgage decisions, and the only users of the appraisal that make those decisions are the lenders. That makes every other party in the transaction an off-label users, and every other use of the appraisal an off-label use.
So when a loan officer "uses" an appraisal (and an appraiser) their criteria for what is and isn't acceptable for that use is often in direct conflict with the use that's stated in the report. Same with the borrowers. These parties don't make the decisions towards which the appraisals are ostensibly aimed at, so if/when their interests are of effect on the results it is never aimed at promulgating the legitimate use of the appraisal within the context of its stated usage.
If all the action occurs at the margins it only takes a small percentage of bad appraisals to taint a loan portfolio. And IRL it takes a lot of performing assets to offset the loss from a single non-performing asset, particularly if that asset was grossly overvalued and consequently grossly overencumbered to begin with.