Fix kpar auto-adjustment for PW basis

[chengleizheng]

Reminder

• I have read AGENTS.md and docs/developers_guide/agent_governance.md.
• I have linked an issue or explained why this PR does not need one.
• I have added adequate unit tests and/or case tests, or explained why not.
• I have listed the exact verification commands run and their results.
• I have described user-visible behavior changes, including INPUT parameter changes.
• I have explained core-module impact for ESolver, HSolver, ElecState, Hamilt, Operator, Psi, or other source/ changes.
• I have requested any needed governance exception below.

Linked Issue

Fix #7722

Case Tests for my changes

Commands run:

  cd tests/01_PW/026_PW_KPAR
  echo "device    gpu" >> INPUT
  cd .. 
  bash ../integrate/Autotest.sh -n 1 -a /home/chenglei/abacus-develop/build/abacus_basic_gpu -f <(echo "026_PW_KPAR/")

Result summary:

Before fix:

-----AUTO TESTS OF ABACUS ------
ABACUS path: /home/chenglei/abacus-develop/build/abacus_basic_gpu
Number of processes: 1
Number of threads: 32
Test accuracy totenergy: 0.0000001 eV
Test accuracy force: 0.0001
Test accuracy stress: 0.001
Test accuracy descriptor mean: 0.00001
Check accuaracy: 8
Test cases file: /dev/fd/63
Test cases regex: ^[^#].*_.*$
Generate reference:
--------------------------------

[ RUN      ] 026_PW_KPAR/
--------------------------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
--------------------------------------------------------------------------
--------------------------------------------------------------------------
mpirun detected that one or more processes exited with non-zero status, thus causing
the job to be terminated. The first process to do so was:

  Process name: [[50287,1],0]
  Exit code:    1
--------------------------------------------------------------------------
[ERROR     ] 026_PW_KPAR/ failed.

                     ▄████▄ █████▄ ▄████▄ ▄█████ ██  ██ ▄█████
                     ██▄▄██ ██▄▄██ ██▄▄██ ██     ██  ██ ▀▀▀▄▄▄
                     ██  ██ ██▄▄█▀ ██  ██ ▀█████ ▀████▀ █████▀


                                v3.11.0-beta7

               Atomic-orbital Based Ab-initio Computation at UStc

                     Website: http://abacus.ustc.edu.cn/
               Documentation: https://abacus.deepmodeling.com/
                  Repository: https://github.com/abacusmodeling/abacus-develop
                              https://github.com/deepmodeling/abacus-develop
                      Commit: 7ba39dd18 (Thu Jul 30 13:17:17 2026 +0800)

 Fri Jul 31 17:22:17 2026

 ---------------------------------------------------------
                         !NOTICE!
 ---------------------------------------------------------

 The parameter device has been read twice.
 CHECK IN FILE : warning.log

 For detailed manual of ABACUS, please see the website
 https://abacus.deepmodeling.com
 For any questions, propose issues on the website
 https://github.com/deepmodeling/abacus-develop/issues
 ---------------------------------------------------------
                         !NOTICE!
 ---------------------------------------------------------

 TIME STATISTICS
-----------------------------------------------------
 CLASS_NAME    NAME    TIME/s  CALLS   AVG/s  PER/%
-----------------------------------------------------
            total      0.00   1        0.00   100.00
-----------------------------------------------------

 See output information in :
[----------] Time elapsed: 3.033 seconds

[WARNING]    1 test cases out of 1 failed.
026_PW_KPAR/

After fix:

  -----AUTO TESTS OF ABACUS ------
  ABACUS path: /home/chenglei/abacus-develop/build/abacus_basic_gpu
  Number of processes: 1
  Number of threads: 32
  Test accuracy totenergy: 0.0000001 eV
  Test accuracy force: 0.0001
  Test accuracy stress: 0.001
  Test accuracy descriptor mean: 0.00001
  Check accuaracy: 8
  Test cases file: /dev/fd/63
  Test cases regex: ^[^#].*_.*$
  Generate reference:
  --------------------------------
  
[ RUN      ] 026_PW_KPAR/
  [----------] test parallel method kpar. Must be run by MPI with at least 2 cores, otherwise please ignore this test.
  [      OK  ]  etotref
  [      OK  ]  etotperatomref
  [      OK  ]  pointgroupref
  [      OK  ]  spacegroupref
  [      OK  ]  nksibzref
  [----------] Time elapsed: 3.715 seconds
  
[ PASSED   ]  5 test cases passed.

with warning, runs normally.

Additional context

Although I solve the kpar related problems, 007_PW_UPF201_USPP_Fe, 063_PW_CR and scf_out_chg_tau​ cases still have other issues.

What's changed?

Fixed KPAR auto-adjustment that was broken due to reset_value callback ordering. Previously the check para.input.device == "gpu" in kpar's callback always evaluated to false because device's own callback (which resolves auto→gpu) runs later in the registration order. Removed the #if defined(__CUDA) guard and device == "gpu" check — the auto-adjust is now unconditional for PW basis.

Governance Notes

• INPUT/docs changes: None. KPAR behavior unchanged when NPROC matches; only affects the error case where KPAR > NPROC.
• Core module impact: read_input_item_system.cpp — input parameter post-processing only.
• Exceptions requested: None.