Fits n-dimensional data by means of orthogonal nonlinear least-squares using Levenberg-Marquardt minimization, with a solver that exploits the block structure of the problem as 'ODRPACK' does, so that its cost grows linearly with the number of observations, and provides functionality for fit diagnostics and plotting. Delivers the same results as the 'ODRPACK' Fortran implementation described in Boggs et al. (1989)