One poll alone has a high chance of errors, average a bunch of polls and factor in historical data and you can make very accurate predictions. Nate Silver and his crew over at 538 did that and predicted that Trump had about a 1 in 3 chance of winning. 538 also saw Trump's potential electoral college path to victory in the midwest swing states, Hillary's campaign and the DNC missed that trend and the rest is history.
Would be interesting to do some regression analysis and see if employment was a factor, but given that the drop in unemployment was smaller than the increase in insured, it's unlikely that employment contributed to the entire change. As for Obamacare not being responsible, it's the major variable that changed during the period studied, and no other variable can account for the size of the change in insured percentage.
As for Obamacare failing, depends how you define "failing". Overall, Obamacare is succeeding in most measures like insured rate. And while a few folks may have fallen through the cracks and are being gouged by the insurance companies, millions now have health care because of Obamacare.
Sad to see you bashing the good folks (many veterans) who work for and run the VA, but the VA is a model of successfully providing health care while controlling costs. This is largely because unlike Medicare, the VA is allowed to negotiate for the best deals on drugs, DME, etc.. The VA's problems are largely due to surges in patient loads thanks to the Afghanistan and Iraq wars and the tendency of older vets to migrate to warm places like Florida and Arizona.