Vacuum Chamber Design
I've done large high velocity process air ducts that pull a vacuum, although not quite that much. One detail that is commonly used is the exact one you mentioned - welding angles into the corners. The way it's normally done is to leave the outside point of the angle visible from the outside. This allows the plate to be filleted to the angle from the outside and the angle to be filleted to the plate from the inside. As you found out, the problem with bringing two plates together at the corners is cracking from the fatigue of pressurization cycles. It is at the corners where the maximum moment usually occurs.
As for your particular case with the floor, I don't understand why the floor would be the problem if the details are the same all the way around. It seems that under a vacuum, the weight of the floor would help the situation while the roof's weight would be more problematic coupled with the vacuum. Once you add 24,000 lbs evenly distributed, I just can't see why the floor would curve in.