Memory leak in OperatorEXXPW when ik > nks triggers early return

[mohanchen]

Describe the bug

Summary

get_exx_potential and get_exx_stress_potential in source/source_pw/module_pwdft/op_pw_exx_pot.cpp allocate a CPU buffer (pot_cpu = new Real[npw]) but can return early or throw before the matching delete[], leaking the allocation.

Affected locations

Function	new line	Leaking exit	delete[] line
get_exx_potential	37	return at line 52	225
get_exx_potential	37	throw at line 213	225
get_exx_stress_potential	250	throw at line 403	415

• Early return: if (ik > nks) { return; } at line 52 leaks pot_cpu every time it triggers.
• GPU exception path: If cudaHostRegister fails, the throw at lines 213 / 403 leaks pot_cpu and also leaves the pointer registered.

Impact

• Leak size: sizeof(Real) * npw bytes per call (typically ~100 KB–1 MB).
• Trigger frequency: Any EXX calculation where ik > nks (e.g., certain k-parallel or restart paths). If this happens once per SCF step per k-point, the leak is small but unbounded over long runs.

Suggested fix

Move the allocation after the early-return guard, or wrap pot_cpu in a std::unique_ptr<Real[]>:

### 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).