On pytest < 5.3, `lazy_value` parameters are wrongly inserted into `module_results_df` as integer instead of object

[smarie]

With pytest<5.3 when collecting results from pytest cases, the parameters appear as "111111" in module_results_df instead of their actual case name.

Cause: unfortunately on pytest < 5.3, pytest-cases lazy_values inherit from int, so as to have their id correctly created by the pytest id maker engine from their __str__ method. See https://smarie.github.io/python-pytest-cases/pytest_goodies/#parametrize

Workaround:

• we could in pytest-harvest get_filtered_results_df wrap such lazy values (detected using is_lazy_value from pytest case) and wrap them using wrapt.ObjectProxy

• an alternative would be to separate such lazy value instances from the other columns and give them an explicit object dtype, while letting pandas use the auto-detection for the rest.

• last option would be in get_session_synthesis_dct to detect such values and "resolve" them

[smarie]

Finally went for the last option, levaraging new pytest-cases 2.3.0