The fatigue (hard spot) problem can be solved, and is covered in various specs that require a soldered splice to be firmly affixed to something else, i.e. with an attachment to the bulkhead or firmly taped into a wire harness so it is supported and can't flex at the solder joint.
With a crimp...I still like to do the same thing, make sure the xtra mass of the splice is attached to SOMEthing so it can't work the wires.
But on the heat...for routine connections if you can generate that much heat, you're in trouble anyway. A fuse should blow long before the solder can melt. For *cable* lugs, which are carrying high current normally and can carry 1000-4000+ amps during a short, I'll agree with you that solder melting and releasing a connection is more of a real concern. Oddly enough....in the elevator industry, where they use "the same" heavy cables and lugs, the standard used to be SOLDER into the lugs. I've used "elevator solder" and have no idea why it was supposed to be special, just proves that it takes all types, I suppose.
Personally, I swage battery lug fittings--but only after I've placed solder paste ("Solder-It") inside the cable. After the swage is made, I heat the lug enough so the solder paste flows and the resin bleeds out. I'm sure that's "wrong" but in my book, it guarantees the physical crimp has been made, AND an airtight solder connection is in there as well to exclude moisture. I guess in ten or twenty more years I'll find out just how well that's been aging.