Residual error for cacqr with complete_inv=0

[huttered40]

Its reporting that the residual error is large when running a single iteration cacqr with complete_inv=0, meaning that we do the solve. The solve attains the Q factor, and the deviation from orthogonality is fine. Whats interesting is that the residual error with 2 iterations is fine.

I looked for an obvious bug and can't find one, and when I print the Q,R factors out there are no obvious errors by inspection.