Allow creation of Metric from data structure without __name__

[cmd-ntrf]

When applying a function to a time series in a query to Prometheus, the Prometheus server returns a data structure without __name__ key. Apart from that aspect, the time series could still be used to create a Metric object.

This PR proposes to set "metric_name" to None when __name__ is missing.

[Copilot]

Pull request overview

This PR modifies the Metric class initialization to support creating Metric objects from Prometheus query results that lack a __name__ key, which commonly occurs when applying functions to time series. The change sets metric_name to None when __name__ is absent, rather than raising a KeyError.

Key Changes

• Modified line 57 to use pop("__name__", None) instead of direct dictionary access
• Removed the explicit deletion of __name__ from label_config on the previously existing line (now removed), relying on pop() to handle removal

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The PR allows metric_name to be None, but there's no test coverage for this new behavior. Consider adding a test case that verifies:

1. A Metric can be created from data without a __name__ key
2. The metric_name attribute is correctly set to None
3. The label_config contains all other labels
4. Operations like __eq__ and __str__ work correctly with None metric_name

Example test:

def test_metric_without_name(self):
    """Test creating a Metric from data without __name__ key."""
    test_metric = Metric(
        metric={
            "metric": {"instance": "localhost", "job": "test"},
            "value": [1627485628.789, "26.82"],
        }
    )
    self.assertIsNone(test_metric.metric_name)
    self.assertEqual(test_metric.label_config, {"instance": "localhost", "job": "test"})

[4n4nd]

@copilot open a new pull request to apply changes based on the comments in this thread