VNL::vkb is allocated in CPU even if device set to gpu, resulting in bad_array_new_length

[Cstandardlib]

Describe the bug

The vkb memory block of size (nkb, npwx) is allocated by new (CPU MEMORY),
and this snippet is called even in GPU branch.

in source/source_pw/module_pwdft/vnl_pw.cpp:

if (nkb > 0 && allocate_vkb)
    {
        vkb.create(nkb, npwx);
        ModuleBase::Memory::record("VNL::vkb", nkb * npwx * sizeof(std::complex<double>));
    }

in source/source_base/complexmatrix.cpp:

// reallocate memory for Complex Matrix
void ComplexMatrix::create(const int nr_in, const int nc_in, const bool flag_zero)
{
	if( nr_in && nc_in )
	{
		if(c)
		{
			const int size_in=nr_in*nc_in;
			if( size_in!=nr*nc )
			{
				delete[] c;
				c = new std::complex<double>[size_in];
			}
		}
		else
		{
			c = new std::complex<double>[nr_in * nc_in];
		}

		nr = nr_in;
		nc = nc_in;
		size = nr*nc;
		if(flag_zero)	zero_out();
	}
	else
	{
		if(c)	delete[] c;
		c = nullptr;
		nr = nr_in;
		nc = nc_in;
		size = nr*nc;
	}
}

When running on GPU-like device with many atoms and huge ecutwfc, the new expression will raise a std::bad_array_new_length error.

Expected behavior

No response

To Reproduce

No response

Environment

No response

Additional Context

No response

Task list for Issue attackers (only for developers)

• Verify the issue is not a duplicate.
• Describe the bug.
• Steps to reproduce.
• Expected behavior.
• Error message.
• Environment details.
• Additional context.
• Assign a priority level (low, medium, high, urgent).
• Assign the issue to a team member.
• Label the issue with relevant tags.
• Identify possible related issues.
• Create a unit test or automated test to reproduce the bug (if applicable).
• Fix the bug.
• Test the fix.
• Update documentation (if necessary).
• Close the issue and inform the reporter (if applicable).