llama : add log about loading model tensors

[ggerganov]

Minor QoL improvement - sometimes it can be confusing why it takes so long to start processing. We now print a message to explain that the model data is being loaded.

...
0.00.189.951 I print_info: EOG token        = 128001 '<|end_of_text|>'
0.00.189.951 I print_info: EOG token        = 128009 '<|eot_id|>'
0.00.189.952 I print_info: max token length = 256
0.00.189.952 I load_tensors: loading model tensors, this can take a while... (mmap = true)
...