Denis Kuznedelev

Added support for the recent Phi-3 model.

This PR raises asserts when conditioning is not provided to AdaLayerNorm. Specifically: 1) One has to provide either timestep or temb argument to `AdaLayerNorm`. 2) Modulation embedding has to be...

Script `models/llama4/scripts/quantize.py` imports from [non-existing](https://github.com/meta-llama/llama-models/blob/eececc27d275a0f7d4d14ec83648d51c10a76560/models/llama4/scripts/quantize.py#L20) `quantize_impls.py` which was likely supposed to be present in `quantization` directory. I think it is supposed to point [here](https://github.com/meta-llama/llama-models/blob/1ba7d45b4f6f3f89feed1e6fae8a0c4b8cc2932f/models/quantize_impls.py#L74).