lmp: fix_dplr use the same type_map from pair_deepmd

[njzjz]

When pair_coeff has set LAMMPS type_map (e.g. pair_coeff * * O H), fix_dplr uses the same LAMMPS type_map (i.e. O H in this case) to map from LAMMPS types to the DP model types, considering LAMMPS types should always be the same among different commands.

[njzjz]

@Yi-FanLi, I am unsure whether test_pair_deepmd_lr_type_map has covered all the lines I modified.

[codecov]

Codecov Report

Patch coverage: 78.18% and project coverage change: +0.02% 🎉

Comparison is base (f97b4f2) 78.72% compared to head (df7f8fd) 78.75%.
Report is 5 commits behind head on devel.

Additional details and impacted files

@@            Coverage Diff             @@
##            devel    #2776      +/-   ##
==========================================
+ Coverage   78.72%   78.75%   +0.02%     
==========================================
  Files         241      242       +1     
  Lines       25610    25686      +76     
  Branches     1517     1526       +9     
==========================================
+ Hits        20162    20228      +66     
- Misses       5058     5062       +4     
- Partials      390      396       +6     

Files Changed	Coverage Δ
source/api_cc/include/DeepTensor.h	87.50% <ø> (ø)
source/lmp/fix_dplr.cpp	74.36% <71.42%> (-0.23%)	⬇️
source/api_c/include/deepmd.hpp	85.80% <100.00%> (+0.11%)	⬆️
source/api_c/src/c_api.cc	80.81% <100.00%> (+0.19%)	⬆️
source/api_cc/src/DeepTensor.cc	78.54% <100.00%> (+0.15%)	⬆️
source/lmp/pair_deepmd.cpp	53.51% <100.00%> (+0.11%)	⬆️

... and 7 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Yi-FanLi]

I reviewed this pull request. Most of them are fine. The test works as how it is designed. The only thing I want to ask is the question which I raised below.