fix manybody kernel etype issue

[nw13slx]

As mentioned in #165, the multi-element many body kernel should use an independent list for the atomic species in bond_array_mb instead of relying on the etypes for bond_array_2b .

New attribute bond_array_mb_etypes are added to the AtomicEnvironment and the arguments passed to the jit function in mc_simple.py and mc_sephyps.py are slightly modified to get this.

New unit tests for mc_sephyps.py in another branch 2b8e90d will be pulled later.

Close #165

[codecov-io]

Codecov Report

Merging #166 into development will increase coverage by 1.76%.
The diff coverage is 80.00%.

@@               Coverage Diff               @@
##           development     #166      +/-   ##
===============================================
+ Coverage        57.42%   59.19%   +1.76%     
===============================================
  Files               35       35              
  Lines             7392     7894     +502     
===============================================
+ Hits              4245     4673     +428     
- Misses            3147     3221      +74     

Impacted Files	Coverage Δ
flare/kernels/sc.py	26.94% <ø> (+2.27%)	⬆️
flare/env.py	30.73% <33.33%> (+<0.01%)	⬆️
flare/kernels/mc_sephyps.py	18.49% <100.00%> (ø)
flare/kernels/mc_simple.py	33.87% <100.00%> (ø)
flare/mgp/mgp_en.py	82.66% <0.00%> (-1.88%)	⬇️
flare/kernels/utils.py	95.40% <0.00%> (ø)
flare/gp_from_aimd.py	91.07% <0.00%> (+1.82%)	⬆️
flare/gp.py	81.81% <0.00%> (+4.65%)	⬆️
... and 2 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 fcee660...7bdeb6d. Read the comment docs.