[BUG] Dual simplex produces wrong sign for Lagrange multiplers for >= constraints

[chris-maes]

We received a report from a user that dual simplex is producing the wrong sign for lagrange multipliers associated with rows in the form a_i^T x >= b.

Internally, these rows are first converted to a_i^T x <= b and then slacks are added. But it looks like we don't adjust the Lagrange multipliers after the solve.