Add support for Nomic Embed

[cebtenzzre]

Nomic Embed is now working in llama.cpp using these changes (f32 MSE is 3.6e-10 on a simple test).

This PR also changes the llama.cpp names of some of BERT's tensors to match the more accurate ones the conversion script is currently using.

Note: This pull request is currently built on top of #5466, so that should be merged first.

[jacoobes]

ur such a goat ( greatest of all time)

[iamlemec]

Awesome! I had to do the following to convert-hf-to-gguf.py to get it to run but after that it worked like a charm and matches up with sentence_transformers.

     def set_gguf_parameters(self):
         super().set_gguf_parameters()
-        self.gguf_writer.add_causal_attention(self.hparams["causal"])
+        self.gguf_writer.add_context_length(self.hparams["n_ctx"])
         self.gguf_writer.add_rope_freq_base(self.hparams["rotary_emb_base"])
-        self.gguf_writer.add_pooling_layer(True)
+        self.gguf_writer.add_layer_norm_eps(self.hparams["layer_norm_epsilon"])