paddle.full causes segmentation fault when passed array-like fill value #72709
Description
Summary
当向 paddle.full 函数传递一个数组（array-like）作为 fill_value 参数时，程序发生段错误（Segmentation Fault）。此行为属于用户误用，但框架抛出合适的异常信息是不是比直接崩溃更为合适。
Code to reproduce bug
import paddle
import numpy as np
np.random.seed(42)
# This line triggers the segfault
input_tensor = paddle.full([1, 1], np.random.randn(1, 1) + 1j, dtype='complex64')Output:
--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
0 paddle::pybind::eager_api_full(_object*, _object*, _object*)
1 paddle::pybind::CastPyArg2Scalar(_object*, std::string const&, long)
2 paddle::pybind::CastNumpy2Scalar(_object*, std::string const&, long)
3 paddle::pybind::PyObject_ToDouble(_object*)
----------------------
Error Message Summary:
----------------------
FatalError: `Segmentation fault` is detected by the operating system.
[TimeInfo: *** Aborted at 1747198854 (unix time) try "date -d @1747198854" if you are using GNU date ***]
[SignalInfo: *** SIGSEGV (@0x0) received by PID 876 (TID 0x7f4cfa501740) from PID 0 ***]
Segmentation fault (core dumped)
System Information
OneFlow installation: pip
OS: Ubuntu 20.04
OneFlow version (run python3 -m oneflow --doctor):
path: ['/root/miniconda3/envs/myconda/lib/python3.8/site-packages/oneflow']
version: 0.9.0
git_commit: 381b12c
cmake_build_type: Release
rdma: True
mlir: True
Python version: 3.8
CUDA driver version: 11.6
GPU models: NVIDIA A16
其他补充信息 Additional Supplementary Information
No response