Implemented a temporary patch for recently broken MiniMax-M2 (in builds after 12/10) for some quants.

Dockerfile

@@ -80,6 +80,9 @@ RUN python3 use_existing_torch.py && \
 COPY fastsafetensors.patch .
 RUN patch -p1 < fastsafetensors.patch
 
+# TEMPORARY PATCH for broken MiniMax M2 - tracking https://github.com/vllm-project/vllm/issues/30445 and https://github.com/vllm-project/vllm/pull/30389
+RUN curl -L https://patch-diff.githubusercontent.com/raw/vllm-project/vllm/pull/30389.diff | git apply
+
 # Final Build
 # Uses --no-build-isolation to respect the pre-installed Torch/FlashInfer
 RUN pip install --no-build-isolation . -v