fix: avoid copy constructor instantiation in shared_ptr fallback cast (#6028)

* tests: add regressions for shared_ptr reference_internal fallback

* fix: avoid copy constructor instantiation in shared_ptr fallback cast

* Remove stray empty line

* tests: rename PyTorch shared_ptr regression test files

* refactor: add cast_non_owning helper for reference-like casts

Name the non-owning generic cast path so callers do not have to rediscover that
reference-like policies must pass null copy/move constructor callbacks. This
keeps the shared_ptr reference_internal fallback self-documenting and points
future maintainers toward the safe API.

Made-with: Cursor

* tests: guard deprecated-copy warning probes with __has_warning

Use __has_warning for the Clang-only regression test so older compiler jobs skip
unsupported warning groups instead of failing with -Wunknown-warning-option. A
simple __clang_major__ >= 13 guard would be shorter, but it bakes in a version
cutoff; __has_warning is slightly more verbose while being more robust to
vendor builds, backports, and future packaging differences.

Made-with: Cursor

---------

Co-authored-by: Ralf W. Grosse-Kunstleve <rgrossekunst@nvidia.com>

tests/test_pytorch_shared_ptr_cast_regression.cpp

@@ -0,0 +1,62 @@
+#include "pybind11_tests.h"
+
+#include <memory>
+#include <string>
+
+#if defined(__clang__)
+#    if __has_warning("-Wdeprecated-copy-with-user-provided-dtor")
+#        pragma clang diagnostic error "-Wdeprecated-copy-with-user-provided-dtor"
+#    endif
+#    if __has_warning("-Wdeprecated-copy-with-dtor")
+#        pragma clang diagnostic error "-Wdeprecated-copy-with-dtor"
+#    endif
+#endif
+
+namespace test_pytorch_regressions {
+
+// Directly extracted from PyTorch patterns that regressed in CI.
+struct TracingState : std::enable_shared_from_this<TracingState> {
+    TracingState() = default;
+    ~TracingState() = default;
+    int value = 0;
+};
+
+const std::shared_ptr<TracingState> &get_tracing_state() {
+    static std::shared_ptr<TracingState> state = std::make_shared<TracingState>();
+    return state;
+}
+
+struct InterfaceType {
+    ~InterfaceType() = default;
+    int value = 0;
+};
+using InterfaceTypePtr = std::shared_ptr<InterfaceType>;
+
+struct CompilationUnit {
+    InterfaceTypePtr iface = std::make_shared<InterfaceType>();
+
+    InterfaceTypePtr get_interface(const std::string &) const { return iface; }
+};
+
+} // namespace test_pytorch_regressions
+
+TEST_SUBMODULE(pybind11_pytorch_regressions, m) {
+    using namespace test_pytorch_regressions;
+
+    py::class_<TracingState, std::shared_ptr<TracingState>>(m, "TracingState")
+        .def(py::init<>())
+        .def_readwrite("value", &TracingState::value);
+
+    m.def("_get_tracing_state", []() { return get_tracing_state(); });
+
+    py::class_<InterfaceType, InterfaceTypePtr>(m, "InterfaceType")
+        .def(py::init<>())
+        .def_readwrite("value", &InterfaceType::value);
+
+    py::class_<CompilationUnit, std::shared_ptr<CompilationUnit>>(m, "CompilationUnit")
+        .def(py::init<>())
+        .def("get_interface",
+             [](const std::shared_ptr<CompilationUnit> &self, const std::string &name) {
+                 return self->get_interface(name);
+             });
+}