supports_constraint can be wrong

[odow]

x-ref #2865 (comment)

Check this out:

julia> import Ipopt; import MathOptInterface as MOI

julia> model = MOI.instantiate(Ipopt.Optimizer; with_bridge_type=Float64)
MOIB.LazyBridgeOptimizer{IpoptMOIExt.Optimizer}
├ Variable bridges: none
├ Constraint bridges: none
├ Objective bridges: none
└ model: IpoptMOIExt.Optimizer
  ├ ObjectiveSense: FEASIBILITY_SENSE
  ├ ObjectiveFunctionType: MOI.ScalarAffineFunction{Float64}
  ├ NumberOfVariables: 0
  └ NumberOfConstraints: 0

julia> MOI.supports_constraint(
           model,
           MOI.VectorOfVariables,
           MOI.Indicator{MOI.ACTIVATE_ON_ONE,MOI.EqualTo{Float64}},
       )
true

This is wrong, because it assumes that the variable will be binary, which is not part of the constraint.

[odow]

Maybe the IndicatorToMILPBridge should return true iff the model supports ZeroOne.