fix(exx): declare explicit specializations before extern template instantiation

The extern template class declarations in op_pw_exx.h trigger implicit
instantiation of all template members when the header is included. The
explicit specializations of cal_density_recip and rho_recip2real in the
.cpp file were therefore "after instantiation", causing a compile error.

Fix by adding forward declarations of all 8 explicit specializations
(CPU + GPU, cal_density_recip + rho_recip2real, complex<double> +
complex<float>) in the header, placed before the extern template block.

Author: abacus_fixer

source/source_pw/module_pwdft/op_pw_exx.h

@@ -175,6 +175,63 @@ class OperatorEXXPW : public OperatorPW<T, Device>
 
 };
 
+// Explicit specializations must be declared before any implicit instantiation.
+// The extern template declarations below would otherwise instantiate the
+// generic cal_density_recip / rho_recip2real members.
+template <>
+void OperatorEXXPW<std::complex<double>, base_device::DEVICE_CPU>::cal_density_recip(
+    const std::complex<double>* psi_nk_real,
+    const std::complex<double>* psi_mq_real,
+    double omega) const;
+
+template <>
+void OperatorEXXPW<std::complex<float>, base_device::DEVICE_CPU>::cal_density_recip(
+    const std::complex<float>* psi_nk_real,
+    const std::complex<float>* psi_mq_real,
+    double omega) const;
+
+template <>
+void OperatorEXXPW<std::complex<double>, base_device::DEVICE_CPU>::rho_recip2real(
+    const std::complex<double>* rho_recip,
+    std::complex<double>* rho_real,
+    bool add,
+    double factor) const;
+
+template <>
+void OperatorEXXPW<std::complex<float>, base_device::DEVICE_CPU>::rho_recip2real(
+    const std::complex<float>* rho_recip,
+    std::complex<float>* rho_real,
+    bool add,
+    float factor) const;
+
+#if ((defined __CUDA) || (defined __ROCM))
+template <>
+void OperatorEXXPW<std::complex<double>, base_device::DEVICE_GPU>::cal_density_recip(
+    const std::complex<double>* psi_nk_real,
+    const std::complex<double>* psi_mq_real,
+    double omega) const;
+
+template <>
+void OperatorEXXPW<std::complex<float>, base_device::DEVICE_GPU>::cal_density_recip(
+    const std::complex<float>* psi_nk_real,
+    const std::complex<float>* psi_mq_real,
+    double omega) const;
+
+template <>
+void OperatorEXXPW<std::complex<double>, base_device::DEVICE_GPU>::rho_recip2real(
+    const std::complex<double>* rho_recip,
+    std::complex<double>* rho_real,
+    bool add,
+    double factor) const;
+
+template <>
+void OperatorEXXPW<std::complex<float>, base_device::DEVICE_GPU>::rho_recip2real(
+    const std::complex<float>* rho_recip,
+    std::complex<float>* rho_real,
+    bool add,
+    float factor) const;
+#endif
+
 extern template class OperatorEXXPW<std::complex<float>, base_device::DEVICE_CPU>;
 extern template class OperatorEXXPW<std::complex<double>, base_device::DEVICE_CPU>;
 #if ((defined __CUDA) || (defined __ROCM))