rm 'dp transfer' from the model compression workflow

[denghuilu]

Since in PR #1000 we have enable initialize the fitting-net variables with the frozen model saved variables. So there’s no need for the dp transfer interface to transfer the fitting-net variables to the compressed model.

[codecov-commenter]

Codecov Report

Merging #1008 (0c32e6d) into devel (472d7c4) will decrease coverage by 10.31%.
The diff coverage is n/a.

❗ Current head 0c32e6d differs from pull request most recent head 95b61bc. Consider uploading reports for the commit 95b61bc to get more accurate results

@@             Coverage Diff             @@
##            devel    #1008       +/-   ##
===========================================
- Coverage   74.59%   64.28%   -10.32%     
===========================================
  Files          86        5       -81     
  Lines        6921       14     -6907     
===========================================
- Hits         5163        9     -5154     
+ Misses       1758        5     -1753     

Impacted Files	Coverage Δ
deepmd/fit/polar.py
deepmd/infer/deep_pot.py
deepmd/utils/learning_rate.py
deepmd/utils/neighbor_stat.py
deepmd/loggers/loggers.py
deepmd/utils/data_system.py
deepmd/loss/__init__.py
source/op/_prod_virial_se_r_grad.py
deepmd/common.py
source/op/_prod_force_grad.py
... and 68 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 472d7c4...95b61bc. Read the comment docs.