> another day or two
At this point, due to the moon's orbit interfering with the trajectory of this launch, they have to wait another week or two, not a day, if they scrub this evening's launch. Waiting 1-2 more week for this launch might mean pushing other customer's launch dates back even further, thus incurring more delays and more cost overruns won't help SpaceX improve anything faster.
Not trying to be snarky - I have no idea! But that could be reason?
Imagine if you paid UPS to send a package across the country and they delayed your package by a week because they're testing a new cargo plane and conditions aren't currently favorable for testing, even though the normal cargo plane would do the job fine right now. You'd probably be pretty unhappy that UPS is putting their own internal testing ahead of doing what you paid them for.
Yeah, due to weather and technical problems. Obviously there's no way of getting around those. It would be a much bigger deal to the customer to delay just because "SpaceX wants to".
> how could delaying one more day be worth more (ie cost to SpaceX) than the value of the recovered rocket times the probability of recovery. And the correct answer is ... it would be a delay of over a week, not just another day
No, even if they had a launch window tomorrow, they would not have delayed another day. The correct answer is that the price of delaying just to conduct an optional test is decreased confidence from potential launch customers that their payloads will not be delayed due to SpaceX's whims, customers who already question SpaceX's ability to launch on time. When someone pays you tens of millions of dollars, you don't get to delay their flight just because you want to test something experimental - it doesn't work that way. SpaceX will have many more opportunities to test their landing procedures, it's not worth pissing off a customer just to be more aggressive on the recovery timeline.