MEGNET model and evaluate the design

[VIGNESHinZONE]

MEGNET- Transfer learning model for Crystal Graphs and might be some of the important models for deepchem's future plans in adding model-hub.
Currently, in this PR, I have built a wrapper function of the existing MEGNET model suitable with deepchem's KerasModel API.

I will soon add the test functions and documentation but for now, we need to evaluate the design choice of the wrapper function.

MEGNET uses BSD-3-Cluase-License.

[codecov-commenter]

Codecov Report

Merging #2485 (c001017) into master (341cdf9) will increase coverage by 0.01%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #2485      +/-   ##
==========================================
+ Coverage   85.35%   85.36%   +0.01%     
==========================================
  Files         307      309       +2     
  Lines       27048    27287     +239     
==========================================
+ Hits        23086    23293     +207     
- Misses       3962     3994      +32     

Impacted Files	Coverage Δ
...hem/feat/material_featurizers/megnet_featurizer.py	0.00% <0.00%> (ø)
deepchem/models/megnet.py	0.00% <0.00%> (ø)
...m/feat/molecule_featurizers/mol2vec_fingerprint.py	90.32% <0.00%> (-3.23%)	⬇️
deepchem/models/chemnet_models.py	78.32% <0.00%> (-0.30%)	⬇️
...em/feat/molecule_featurizers/one_hot_featurizer.py	86.00% <0.00%> (-0.12%)	⬇️
deepchem/models/tests/test_reload.py	100.00% <0.00%> (ø)
deepchem/data/tests/test_sdf_loader.py	100.00% <0.00%> (ø)
deepchem/metrics/tests/test_metrics.py	100.00% <0.00%> (ø)
deepchem/feat/tests/test_one_hot_featurizer.py	100.00% <0.00%> (ø)
deepchem/feat/tests/test_smiles_featurizers.py	100.00% <0.00%> (ø)
... and 7 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 341cdf9...c001017. Read the comment docs.