Querying optimizer to set for another JuMP model

[odow]

One possible way I can think is let user store the solver as solver = GLPK.Optimizer before the first model is run, and then simply use solver in my second model.

This is the recommended way, just ask the user for an optimizer.