feat: add metric querying and filtering to Series and Experiment

[tejapulagam]

Add metric querying and filtering

Users need to slice into logged metrics programmatically, ex: filter by name pattern, value range, or index, and get aggregate stats without pulling raw data.

# glob-match + value filter
exp.query_metrics(name="train/*", max_value=0.5)

# quick stats without touching raw values
exp.query_metrics(name="*/loss", summary_only=True)

# series-level filtering
exp["loss"].filter(min_value=0.01, start_index=100)
exp["loss"].summary()  # → MetricSummary(count, min, max, mean, std, median, ...)