Updated training window error assertion for relative units

[christopherbunn]

• Updated the error assertion for query_by_values so that an error is asserted only when the training window is in observations Why does training_window in dfs() not support 'month'? #719
• Added a test case for when a relative unit is used for the training window ft.dfs training_window doesn't accept Timedelta #618
• Updated test_calcuate_feature_matrix.py for new error assertion message

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@3951bfe). Click here to learn what that means.
The diff coverage is 100%.

@@            Coverage Diff            @@
##             master     #728   +/-   ##
=========================================
  Coverage          ?   97.64%           
=========================================
  Files             ?      118           
  Lines             ?    10247           
  Branches          ?        0           
=========================================
  Hits              ?    10006           
  Misses            ?      241           
  Partials          ?        0

Impacted Files	Coverage Δ
featuretools/synthesis/dfs.py	100% <ø> (ø)
featuretools/entityset/entity.py	96.09% <100%> (ø)
...utational_backend/test_calculate_feature_matrix.py	99.35% <100%> (ø)
featuretools/tests/synthesis/test_dfs_method.py	100% <100%> (ø)

---

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 3951bfe...b6bb5e6. Read the comment docs.

[jeremyliweishih]

LGTM. What do you think @rwedge?

[jeremyliweishih]

Would it be possible to add testing for edge cases like leap years? Otherwise LGTM!

[jeremyliweishih]

LGTM. @rwedge

[rwedge]

I don't think these assertions are working as intended. assert feature_matrix_4.index.all([2, 3, 4]) and assert feature_matrix_4.index.all([2, 3]) are both evaluating as True

[rwedge]

Looks good!