implement getsolvetime. issue #405

[odow]

Exposes MathProgBase.getsolvetime and checks if implemented by solver. #405

[mlubin]

LGTM. The only question is if we should follow the MPB naming convention here, given that we decided to do so for numvar etc (#356). I'd lean towards extending the MPB method instead of defining a new name if it just accesses or imitates a property of the MPB solver.

[mlubin]

I think it's more Julia-like (for type stability reasons) to throw an exception when the value isn't available instead of returning nothing.

[mlubin]

@IainNZ @joehuchette, comments on the naming issue?

[IainNZ]

I'm leaning towards MathProgBase version, for now.

[joehuchette]

I think I agree with Iain in theory, except for the fact that the current way to call it is so verbose with getInternalSolver.

[mlubin]

@joehuchette, this wrapper lets you call it without getInternalSolver

[joehuchette]

Understood, I'm thinking that I would just prefer a shorter alias for getInternalSolver

[mlubin]

@joehuchette, the wrapper method lets you give a nice error message when the solver doesn't implement getsolvetime, which is pretty useful.

[mlubin]

@odow, we can merge this as is. Could you squash the commits?

[odow]

@mlubin squashed.

[mlubin]

Thanks!