Why does IPOPT still give a solution after a termination status of local infeasibility, numerical error or iteration limit?

Question

I am solving a NLP in IPOPT and then saving the solution afterwards. I noticed that even when I get a bad termination status of local infeasibility, numerical error or iteration limit IPOPT still returns a solution. Are these solutions useful?

For more context here is what my solution looks like saved as a json:

{"solve_time":5315.037714958191,"optimizer":"Ipopt","termination_status":"NUMERICAL_ERROR","dual_status":"UNKNOWN_RESULT_STATUS","primal_status":"UNKNOWN_RESULT_STATUS","objective":2.386034577709679e7,"solution":{ x =
... very long solution }