While not a total disaster, it is not a fixable problem; the frame is too stout and too large to attempt straightening it. The problem with an out-of-square frame is that the rest of the house must be built on top of the frame, and it's important that the house itself be square. Building a square house on a non-square frame is definitely doable, but it would've been significantly easier had the trailer frame been squarer.
I made some inquiries as to how the error occurred. Apparently, in drawing up the specification for the trailer frame, the team had simply not thought to specify a squareness. It would've taken seconds to add that specification. Doing so might have avoided many hours of re-design or re-work. Whoever wrote up the specification seems to have done a fine job otherwise. They simply didn't think of specifying a squareness. It was an unknown unknown.
Call it the Donald Rumsfeld problem. As Rummy famously said in discussing WMDs in Iraq,
In WMDs and in developing prototypes, even just one unknown unknown can be a doozy.
The trick with prototype development is to maximize the known knowns and minimize the unknown unknowns. Easier said than done. Every time I go through the process, I get a little better at spotting the unknowns.
In my current capacity as Resident Metalworking Geek, I'm sometimes approached by colleagues and students about building things. The first conversation is often an interesting one. Often the "client" is thinking of all the exciting possibilities, while I'm simultaneously trying to anticipate everything that might go wrong. I'm not trying to crush dreams (although it probably feels that way); I'm trying to keep the dream alive long-term by anticipating the unknowns.