[Bugfix]Remove conflicting triton after vllm-ascend install on x86 (#7497)

### What this PR does / why we need it?
This PR fixes the x86 image issue where both `triton` and
`triton-ascend` are installed in the final environment.
- https://github.com/vllm-project/vllm-ascend/issues/7359

We confirmed the root cause is not that `triton` fails to uninstall
after the upstream `vllm` installation. Instead, during the
`vllm-ascend` installation step, pip resolves and installs upstream
`triton` again alongside `triton-ascend` on x86 platforms. This leads to
module conflicts at runtime because both distributions provide the
`triton` Python package.

To fix this, this PR updates all Dockerfiles to remove upstream `triton`
immediately after installing `vllm-ascend`, while keeping the
`triton-ascend` version resolved by `vllm-ascend` itself.

Affected files:
- `Dockerfile`
- `Dockerfile.a3`
- `Dockerfile.310p`
- `Dockerfile.openEuler`
- `Dockerfile.a3.openEuler`
- `Dockerfile.310p.openEuler`

### Does this PR introduce _any_ user-facing change?
Yes.

For x86 container images, the final Python environment will no longer
keep upstream `triton` alongside `triton-ascend`. This avoids importing
the wrong Triton package and fixes related runtime failures.

### How was this patch tested?
Root cause validation was performed by reproducing the installation flow
locally and checking the package state after each step.

Observed during `vllm-ascend` installation on x86:
- `triton-ascend` was installed as expected
- upstream `triton` was also installed again in the same step
``` bash
export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
source /usr/local/Ascend/nnal/atb/set_env.sh && \
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
python3 -m pip cache purge

Successfully installed aiofiles-25.1.0 arctic-inference-0.1.1 blinker-1.9.0 cmake-4.2.3 fastapi-0.123.10 
flask-3.1.3 h2-4.3.0 hpack-4.1.0 hypercorn-0.18.0 hyperframe-6.1.0 itsdangerous-2.2.0 numpy-1.26.4 
opencv-python-headless-4.11.0.86 pandas-3.0.1 pandas-stubs-3.0.0.260204 priority-2.0.0 pybind11-3.0.2 
python-dateutil-2.9.0.post0 quart-0.20.0 setuptools-scm-9.2.2 six-1.17.0 starlette-0.50.0 torch-2.9.0+cpu 
torch-npu-2.9.0 torchaudio-2.9.0+cpu torchvision-0.24.0+cpu triton-3.6.0 triton-ascend-3.2.0 
vllm_ascend-0.17.0rc2.dev51+geb92e7d50 werkzeug-3.1.6 wheel-0.46.3 wsproto-1.3.2 xgrammar-0.1.32
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with 
the system package manager, possibly rendering your system unusable. It is recommended to use a virtual 
environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what 
you are doing and want to suppress this warning.
Files removed: 423 (1025.9 MB)
Directories removed: 5
```

- vLLM version: v0.17.0
- vLLM main:
8b6325758c

---------

Signed-off-by: MrZ20 <2609716663@qq.com>

Dockerfile

@@ -59,11 +59,13 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -v -e /vllm-workspace/vllm
 
 # Install vllm-ascend
 # Append `libascend_hal.so` path (devlib) to LD_LIBRARY_PATH
+# Installing vllm-ascend on x86 can pull upstream triton back in alongside triton-ascend. Remove it immediately after this step.
 RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
     source /usr/local/Ascend/nnal/atb/set_env.sh && \
     export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
     python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
+    if [ "$(uname -i)" = "x86_64" ]; then python3 -m pip uninstall -y triton; fi && \
     python3 -m pip cache purge
 
 # Install clang-15 (for triton-ascend)

Dockerfile.310p

@@ -49,11 +49,13 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -v -e /vllm-workspace/vllm
 
 # Install vllm-ascend
 # Append `libascend_hal.so` path (devlib) to LD_LIBRARY_PATH
+# Installing vllm-ascend on x86 can pull upstream triton back in alongside triton-ascend. Remove it immediately after this step.
 RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
     source /usr/local/Ascend/nnal/atb/set_env.sh && \
     export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
     python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
+    if [ "$(uname -i)" = "x86_64" ]; then python3 -m pip uninstall -y triton; fi && \
     python3 -m pip cache purge
 
 # Install modelscope (for fast download) and ray (for multinode)

Dockerfile.310p.openEuler

@@ -44,12 +44,14 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -e /vllm-workspace/vllm/[a
     python3 -m pip cache purge
 
 # Install vllm-ascend
+# Installing vllm-ascend on x86 can pull upstream triton back in alongside triton-ascend. Remove it immediately after this step.
 RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
     source /usr/local/Ascend/nnal/atb/set_env.sh && \
     export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
     export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/include/c++/12:/usr/include/c++/12/`uname -i`-openEuler-linux && \
     python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
+    if [ "$(uname -i)" = "x86_64" ]; then python3 -m pip uninstall -y triton; fi && \
     python3 -m pip cache purge
 
 # Install modelscope (for fast download) and ray (for multinode)

Dockerfile.a3

@@ -58,11 +58,13 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -v -e /vllm-workspace/vllm
 
 # Install vllm-ascend
 # Append `libascend_hal.so` path (devlib) to LD_LIBRARY_PATH
+# Installing vllm-ascend on x86 can pull upstream triton back in alongside triton-ascend. Remove it immediately after this step.
 RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
     source /usr/local/Ascend/nnal/atb/set_env.sh && \
     export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
     python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
+    if [ "$(uname -i)" = "x86_64" ]; then python3 -m pip uninstall -y triton; fi && \
     python3 -m pip cache purge
 
 # Install clang-15 (for triton-ascend)

Dockerfile.a3.openEuler

@@ -58,12 +58,14 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -e /vllm-workspace/vllm/[a
     python3 -m pip cache purge
 
 # Install vllm-ascend
+# Installing vllm-ascend on x86 can pull upstream triton back in alongside triton-ascend. Remove it immediately after this step.
 RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
     source /usr/local/Ascend/nnal/atb/set_env.sh && \
     export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
     export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/include/c++/12:/usr/include/c++/12/`uname -i`-openEuler-linux && \
     python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
+    if [ "$(uname -i)" = "x86_64" ]; then python3 -m pip uninstall -y triton; fi && \
     python3 -m pip cache purge
 
 # Install clang (for triton-ascend)

Dockerfile.openEuler

@@ -58,12 +58,14 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -e /vllm-workspace/vllm/[a
     python3 -m pip cache purge
 
 # Install vllm-ascend
+# Installing vllm-ascend on x86 can pull upstream triton back in alongside triton-ascend. Remove it immediately after this step.
 RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
     source /usr/local/Ascend/nnal/atb/set_env.sh && \
     export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
     export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/include/c++/12:/usr/include/c++/12/`uname -i`-openEuler-linux && \
     python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \
+    if [ "$(uname -i)" = "x86_64" ]; then python3 -m pip uninstall -y triton; fi && \
     python3 -m pip cache purge
 
 # Install clang (for triton-ascend)