Improve performance of aggregation feature calculation

[kmax12]

We had logic that checked for the case where the dataframe we were aggregating on was empty or had no related instances. This logic involved a slow python loop. Due to how we fill in missing values later in the function, this check was unnecessary and could be removed.

In some benchmarks, I saw 25% speeds ups for calculations.

Also, added test for the empty child dataframe case, which was previously untested.

[codecov-io]

Codecov Report

Merging #209 into master will increase coverage by 0.02%.
The diff coverage is 88.88%.

@@            Coverage Diff             @@
##           master     #209      +/-   ##
==========================================
+ Coverage   93.46%   93.48%   +0.02%     
==========================================
  Files          71       71              
  Lines        7784     7757      -27     
==========================================
- Hits         7275     7252      -23     
+ Misses        509      505       -4

Impacted Files	Coverage Δ
...tests/computational_backend/test_pandas_backend.py	100% <100%> (ø)	⬆️
...turetools/computational_backends/pandas_backend.py	91.72% <75%> (+0.43%)	⬆️

---

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 94132fc...141adb5. Read the comment docs.

[rwedge]

Looks good!