testing build linux wheels

pytorch/test-infra#5986 and pytorch/pytorch#141565 changed back to manylinux 2014 for versions of cuda/cpu, testing that this fixes the issue since hte actual error seems to be occuring in the Build the wheel step before hte post script gets run:

https://github.com/pytorch/ao/actions/runs/12041427944/job/33573281510

Author: HDCharles

packaging/post_build_script.sh

@@ -13,7 +13,7 @@ if [[ "$CU_VERSION" == cu* ]]; then
     WHEEL_NAME=$(ls dist/)
 
     pushd dist
-    manylinux_plat=manylinux_2_28_x86_64
+    manylinux_plat=manylinux2014_x86_64
     auditwheel repair --plat "$manylinux_plat" -w . \
     --exclude libtorch.so \
     --exclude libtorch_python.so \