mlx: Gemma4 MTP speculative decoding (#15980)

This change adds support for MTP (multi-token prediction) speculative decoding for the
gemma4 model family.

It includes:
  * support for importing safetensors based gemma4 draft models with `ollama create`
  * a new DRAFT command in the Modelfile for specifying draft models
  * a --quantize-draft flag for the ollama create command to quantize the draft model
  * cache support for speculation
  * changes to the rotating cache to be able to handle MTP correctly
  * sampling support for draft model token prediction

---------

Co-authored-by: Daniel Hiltgen <daniel@ollama.com>

Dockerfile

@@ -212,8 +212,11 @@ COPY --from=cpu dist/lib/ollama /lib/ollama
 COPY --from=build /bin/ollama /bin/ollama
 
 FROM ubuntu:24.04
-RUN apt-get update \
+ARG APT_MIRROR=http://archive.ubuntu.com/ubuntu
+RUN sed -i "s|http://archive.ubuntu.com/ubuntu|$APT_MIRROR|g" /etc/apt/sources.list.d/ubuntu.sources \
+    && apt-get update \
     && apt-get install -y ca-certificates libvulkan1 libopenblas0 \
+    && sed -i "s|$APT_MIRROR|http://archive.ubuntu.com/ubuntu|g" /etc/apt/sources.list.d/ubuntu.sources \
     && apt-get clean \
     && rm -rf /var/lib/apt/lists/*
 COPY --from=archive /bin /usr/bin