Fix: add energy_stable_count to ensure energy threshold is really achieved

[hongriTianqi]

Reminder

• Have you linked an issue with this pull request?
• Have you noticed possible changes of behavior below or in the linked issue?
  if scf_ene_thr is set in INPUT, the energy is converged only when the scf_ene_thr has been fulfilled twice.
• Have you explained the changes of codes in core modules of ESolver, HSolver, ElecState, Hamilt, Operator or Psi? (ignore if not applicable)
  an int stable_energy_count is added in ESolver_FP

Linked Issue

Fix #5011

[QuantumMisaka]

@hongriTianqi Sometimes we can encounter another situation: drho reached 1e-7 while EDIFF only reached 1e-4, can we do something to detect it ?

[hongriTianqi]

@QuantumMisaka You can set scf_ene_thr now to a small value. oh, no. scf_thr and scf_ene_thr are or relationship. In this case, you have to increase scf_thr currently. Maybe you need to raise another issue.

[hongriTianqi]

I have tested this PR. The case in the issue can be solved now:

<< Start SCF iteration.
 ITER     TMAGX      TMAGY      TMAGZ       AMAG        ETOT/eV          EDIFF/eV         DRHO     TIME/s
 DA1      3.73e+00   2.28e-06  -4.25e-01   3.94e+00  -1.37072042e+04   0.00000000e+00   1.0581e+00 110.92
 DA2      3.93e+00  -2.03e-06  -3.44e-01   4.10e+00  -1.37085247e+04  -1.32057816e+00   7.5903e-01  86.72
 DA3      4.02e+00   6.13e-06  -4.77e-01   4.21e+00  -1.37094467e+04  -9.21918555e-01   7.0412e-02  73.85
 DA4      4.19e+00  -1.15e-05  -5.19e-01   4.38e+00  -1.37095188e+04  -7.21822904e-02   6.1596e-03  77.59
 DA5      4.31e+00   6.05e-05  -6.10e-01   4.54e+00  -1.37095291e+04  -1.02978001e-02   2.4583e-03 121.19
 DA6      4.32e+00  -7.12e-05  -5.52e-01   4.56e+00  -1.37095291e+04   5.90551582e-07   1.6169e-03 114.86
 DA7      4.36e+00   8.06e-05  -6.48e-01   4.62e+00  -1.37095329e+04  -3.72805275e-03   5.6930e-04 120.51
 DA8      4.36e+00  -6.43e-05  -5.83e-01   4.61e+00  -1.37095343e+04  -1.37836281e-03   6.8844e-05 124.19
 DA9      4.34e+00   1.17e-04  -6.40e-01   4.60e+00  -1.37095343e+04  -9.78010849e-05   4.0596e-05 154.60
 DA10     4.34e+00  -5.10e-05  -5.96e-01   4.59e+00  -1.37095352e+04  -8.66692188e-04   1.5723e-05 154.82
 DA11     4.34e+00   1.19e-04  -6.05e-01   4.59e+00  -1.37095352e+04   3.23685862e-05   6.7233e-06 174.19
 DA12     4.34e+00  -2.40e-04  -5.89e-01   4.59e+00  -1.37095354e+04  -2.33666558e-04   2.9344e-06 181.47
 DA13     4.34e+00   6.62e-04  -6.07e-01   4.59e+00  -1.37095355e+04  -9.88559794e-05   6.0822e-07 185.89
 DA14     4.34e+00  -5.62e-04  -6.04e-01   4.59e+00  -1.37095357e+04  -1.64176245e-04   1.1672e-07 199.03
 DA15     4.34e+00   1.06e-03  -5.98e-01   4.59e+00  -1.37095357e+04  -1.73131651e-05   1.2087e-07 209.72
 DA16     4.34e+00  -1.36e-03  -5.98e-01   4.59e+00  -1.37095357e+04   2.24533143e-05   1.1272e-07 207.90
 DA17     4.34e+00   7.59e-04  -6.02e-01   4.59e+00  -1.37095357e+04  -4.45663877e-05   2.6705e-08 208.12
 DA18     4.34e+00  -2.01e-03  -6.02e-01   4.59e+00  -1.37095357e+04   2.00607770e-05   6.4388e-08 226.55
 DA19     4.35e+00   8.08e-04  -6.01e-01   4.59e+00  -1.37095357e+04  -1.78516579e-05   1.9266e-08 227.23
 DA20     4.35e+00  -3.63e-04  -6.00e-01   4.59e+00  -1.37095357e+04  -4.39093428e-06   1.7010e-08 222.09
 DA21     4.35e+00  -2.08e-04  -6.03e-01   4.59e+00  -1.37095357e+04  -1.82939792e-06   1.3425e-08 220.43
 DA22     4.35e+00   2.59e-04  -6.02e-01   4.59e+00  -1.37095357e+04   7.47567967e-06   8.3377e-09 219.63
 >> Leave SCF iteration.
 * * * * * *

The corresponding INPUT parameters are:

calculation   scf
ecutwfc 130
scf_thr    1.0e-8
scf_ene_thr    1.0e-6
scf_nmax 100

[WHUweiqingzhou]

@maki49 could you have a look about EXX part?

[WHUweiqingzhou]

This PR will cause all calculations to run an additional SCF step with finite scf_ene_thr. I believe the current strategy comes at too high a cost and requires further discussion.