add_expr is broken

[odow]

See this example. The expression y looks correct, but when used in the objective it substitutes y[i] -> x[i] instead of y[i] -> x[p[I]]:

julia> using ExaModels, NLPModelsIpopt

julia> begin
           core = ExaModels.ExaCore(; concrete = Val(true))
           p = [3, 1, 2]
           x0 = [1, 2, 3]
           core, x = ExaModels.add_var(core, 3; lvar = x0, uvar = x0)
           core, y = ExaModels.add_expr(core, x[p_i] for p_i in p)
           core, _ = ExaModels.add_obj(core, i * y[i] for i in 1:3)
           m = ExaModels.ExaModel(core)
           result = NLPModelsIpopt.ipopt(m; print_level = 0)
           result.objective, sum(i * x0[p[i]] for i in 1:3)
       end
(14.0, 11)

julia> y
Subexpression (reduced)

  s ∈ R^{3}
  s(x,i) = x[i]

julia> y[1], y[2], y[3]
(x[3], x[1], x[2])