LP sensitivity summary doesn't handle nonbasic free variables correctly

[odow]

x-ref #2244 (comment)

julia> model = Model(GLPK.Optimizer)
A JuMP Model
Feasibility problem with:
Variables: 0
Model mode: AUTOMATIC
CachingOptimizer state: EMPTY_OPTIMIZER
Solver name: GLPK

julia> @variable(model, x >= 1)
x

julia> @variable(model, y)
y

julia> @objective(model, Min, x)
x

julia> optimize!(model)

julia> lp_sensitivity_report(model)
ERROR: AssertionError: size(B, 1) == size(B, 2)
Stacktrace:
 [1] lp_sensitivity_report(model::Model; atol::Float64)
   @ JuMP ~/.julia/packages/JuMP/MIPb6/src/lp_sensitivity2.jl:78
 [2] lp_sensitivity_report(model::Model)
   @ JuMP ~/.julia/packages/JuMP/MIPb6/src/lp_sensitivity2.jl:64
 [3] top-level scope
   @ REPL[46]:1

cc @geoffleyland

[odow]

The issue is that we don't have a way to distinguish if a variable is in the basis if it has no bounds:
https://github.com/jump-dev/MathOptInterface.jl/blob/f585f81caddd54b8cc9843c263ee89b28ebc9a9e/src/attributes.jl#L1198-L1199

I guess for now, we can assume that if it has no bounds and is at 0.0, then it is nonbasic, otherwise it is basic. But this doesn't cover super-basic or degenerate problems.