Add eval() to remove Torch warnings during testing

[mellis13]

When running the SmartSim test suite with the Torch backend installed, warnings are surfaced as shown below:

tests/backends/test_dbmodel.py::test_pt_db_model
  /Users/torch/jit/_trace.py:1084: TracerWarning: Output nr 1. of the traced function does not match the corresponding output of the Python function. Detailed error:
  Tensor-likes are not close!
  
  Mismatched elements: 10 / 10 (100.0%)
  Greatest absolute difference: 0.0451509952545166 at index (0, 9) (up to 1e-05 allowed)
  Greatest relative difference: 0.019849935548005467 at index (0, 9) (up to 1e-05 allowed)
    _check_trace(

This appears to stem from not having the eval() function invoked before jit tracing and saving the model. It appears after adding this call that warnings no longer appear.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (b160c05) 90.77% compared to head (e6fe000) 90.72%.
Report is 1 commits behind head on develop.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #472      +/-   ##
===========================================
- Coverage    90.77%   90.72%   -0.06%     
===========================================
  Files           60       60              
  Lines         3837     3837              
===========================================
- Hits          3483     3481       -2     
- Misses         354      356       +2     

see 1 file with indirect coverage changes

[ashao]

Good catch @mellis13, thanks for chasing it down.

[MattToast]

One VERY small comment, but nothing worth holding approval up over. LGTM!! Thanks for squashing those warnings!!

[MattToast]

Incredibly minor detail, but its going to bug me if I don't mention it:

This link is the only one in this list not in sorted order (I know, incredibly pedantic, but my stupid brain needs things to match, hahaha). Can we move it to the bottom?