fix: set explicit dorny base and decouple cython-test skip

Two fixes from code review:

1. Set `base: main` on dorny/paths-filter so backport PRs targeting
   non-default branches still diff against main for path detection.

2. Decouple SKIP_CYTHON_TEST from SKIP_BINDINGS_TEST_OVERRIDE. The
   path-filter override only skips bindings tests; cuda.core Cython
   tests should still run on core-only PRs. Cython skip is now driven
   solely by CTK minor-version mismatch.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: cpcloud

.github/workflows/ci.yml

@@ -111,9 +111,11 @@ jobs:
         if: ${{ startsWith(github.ref_name, 'pull-request/') }}
         uses: dorny/paths-filter@d1c1ffe0248fe513906c8e24db8ea791d46f8590  # v3.0.3
         with:
-          # Compare against the target branch of the PR. When a PR is pushed
-          # to the synthetic pull-request/<N> ref, dorny/paths-filter falls
-          # back to the repository default branch unless overridden.
+          # Explicitly compare against the repository default branch. On push
+          # events dorny/paths-filter defaults to diffing against the prior
+          # commit, which is wrong for backport PRs targeting non-default
+          # branches. Pinning base to main ensures consistent detection.
+          base: main
           filters: |
             bindings:
               - 'cuda_bindings/**'

ci/tools/env-vars

@@ -65,18 +65,20 @@ elif [[ "${1}" == "test" ]]; then
     SKIP_CYTHON_TEST=1
   else
     USE_BACKPORT_BINDINGS=0
+    # Path-filter override only skips bindings tests, NOT cython tests
+    # for other modules (e.g. cuda.core). Cython skip is driven solely
+    # by the build/test CTK minor-version mismatch.
     if [[ "${SKIP_BINDINGS_TEST_OVERRIDE:-0}" == "1" ]]; then
       SKIP_CUDA_BINDINGS_TEST=1
-      SKIP_CYTHON_TEST=1
     else
       SKIP_CUDA_BINDINGS_TEST=0
-      BUILD_CUDA_MINOR="$(cut -d '.' -f 2 <<< ${BUILD_CUDA_VER})"
-      TEST_CUDA_MINOR="$(cut -d '.' -f 2 <<< ${CUDA_VER})"
-      if [[ ${BUILD_CUDA_MINOR} != ${TEST_CUDA_MINOR} ]]; then
-        SKIP_CYTHON_TEST=1
-      else
-        SKIP_CYTHON_TEST=0
-      fi
+    fi
+    BUILD_CUDA_MINOR="$(cut -d '.' -f 2 <<< ${BUILD_CUDA_VER})"
+    TEST_CUDA_MINOR="$(cut -d '.' -f 2 <<< ${CUDA_VER})"
+    if [[ ${BUILD_CUDA_MINOR} != ${TEST_CUDA_MINOR} ]]; then
+      SKIP_CYTHON_TEST=1
+    else
+      SKIP_CYTHON_TEST=0
     fi
   fi
   # We don't test compute-sanitizer on CTK<12 because backporting fixes is too much effort