Issue with ignore_row_if argument in certain expectations

[NehaNRane]

I've encountered a problem with the ignore_row_if argument in three expectations: expect_column_pair_values_a_to_be_greater_than_b, expect_multicolumn_sun_to_equal, and expect_compound_columns_to_be_unique. While the argument works as expected when either_value_is_missing or any_value_is_missing is passed, it behaves unexpectedly when both_values_are_missing / all_values_are_missing or neither / never is used.

Specifically, when using both_values_are_missing / all_values_are_missing or neither / never in the ignore_row_if argument, the following metrics are incorrect:

• unexpected_count
• unexpected_percent
• partial_unexpected_list
• unexpected_percent_total
• unexpected_percent_nonmissing

This issue impacts the accuracy of the expectations and could lead to incorrect interpretations of data quality.

Steps to Reproduce:

1. Use the affected expectations (expect_column_pair_values_a_to_be_greater_than_b, expect_multicolumn_sun_to_equal, expect_compound_columns_to_be_unique).
2. Pass both_values_are_missing / all_values_are_missing or neither / never in the ignore_row_if argument.
3. Observe the incorrect values for the mentioned metrics.

Expected Behavior:
The ignore_row_if argument should accurately handle the specified conditions ( both_values_are_missing / all_values_are_missing or neither / never) and produce correct metrics for the affected expectations.

Additional Context:

• Great Expectations version: 0.18.12
• Python version: 3.9.6
• snowflake-connector-python: 3.8.1
• snowflake-sqlalchemy: 1.5.1