Adaptive Regularisation Methods under Inexact Evaluations for Nonconvex Optimisation and Machine Learning Applications