An Experimental Study on Exploring Strong Lightweight Vision Transformers via Masked Image Modeling Pre-Training