Compute lambda max for a quantile generalized lasso problem.
get_lambda_max(x, y, d, weights = NULL, lp_solver = c("glpk", "gurobi"))This is not exact, but should be close to the exact value of \(\lambda\) such that \(D \hat\beta = 0\) at the solution \(\hat\beta\) of the quantile generalized lasso problem. It is derived from the KKT conditions when \(\tau = 1/2\).