fix(nimbus): skip data points with None bounds in get_max_metric_value

[SAY-5]

Because

• A data point in relative_uplift results can have explicit null (or missing) lower/upper while still being a non-empty dict (e.g. only point is set). The existing if not data_point: continue guard treats these as truthy and proceeds, then abs(None) raises TypeError: bad operand type for abs(): 'NoneType' (Sentry EXPERIMENTER-PROD-16X).

This commit

• Skips the data point when either bound is None before calling abs() in ResultsManager.get_max_metric_value.
• Adds a parameterized test case covering a data point with None lower/upper alongside a sibling branch with valid bounds, asserting the valid bounds still drive the max.

Fixes #15396

[moibra05]

thanks for taking a look at this! changes look simple and enough to fix this issue, nice work 👍

[yashikakhurana]

Hey @SAY-5 thank you for your contribution, I noticed one of the test case is failing

FAILED experimenter/jetstream/tests/test_results_manager.py::TestExperimentResultsManager::test_get_max_metric_value_2 - AssertionError: 0 != 0.7 

would you be able to take a look and fix that please 🙏

[SAY-5]

Fixed the failing test.

[yashikakhurana]

Thank you @SAY-5 looks good 🎉