fix(metrics): GroupBy(ValueDrift) widget keeps decorated display name (#1706)

[jbbqqf]

Summary

Closes #1706.

`ValueDriftCalculation._render` hard-coded the widget title as
`f"Drift in column '{result.column_name}'"` (see
`src/evidently/metrics/column_statistics.py`). When the calculation is
wrapped in `GroupBy(...)`, `GroupByMetricCalculation.calculate` patches
`self.calculation.display_name` so that other metrics (MaxValue, MeanValue,
…) pick up the "group by '<col>' for label: '<label>'" suffix at render
time — they call `self.display_name()` from `StatisticsCalculation.calculate`.
ValueDrift never read the patched name, so its widget always rendered the
bare column name regardless of how it was wrapped, which is exactly the
bug reported in #1706.

Fix

`src/evidently/metrics/column_statistics.py`:

• Pass the GroupBy-decorated `display_name()` through to `_render` only
  when it carries the GroupBy suffix (`" group by '"` substring). This
  preserves the historical "Drift in column 'X'" wording for non-GroupBy
  reports — no surprise behavior change for existing users.
• `_render` now accepts an optional `display_name=` argument and uses it as
  the counter header.

Reproduce BEFORE/AFTER yourself (copy-paste)

```bash
git fetch origin && git fetch https://github.com/jbbqqf/evidently.git fix/1706-groupby-valuedrift-display-name:_pr1706
pip install -q -e ".[dev]" >/dev/null

run_check() {
python - <<'PY'
import numpy as np, pandas as pd
from evidently.core.datasets import Dataset, DataDefinition
from evidently.core.report import Report
from evidently.metrics.column_statistics import ValueDrift
from evidently.metrics.group_by import GroupBy
rng = np.random.default_rng(0)
df = pd.DataFrame({
"value": np.concatenate([rng.normal(size=100), rng.normal(loc=2, size=100)]),
"group": ["a"]*100 + ["b"]*100,
})
ref = Dataset.from_pandas(df, data_definition=DataDefinition())
cur = Dataset.from_pandas(df.copy(), data_definition=DataDefinition())
snap = Report([GroupBy(ValueDrift(column="value"), "group")]).run(cur, ref)
ctx = snap.context
for fp in snap.metric_results:
res = ctx.get_metric_result(fp)
widgets = res.widget if isinstance(res.widget, list) else [res.widget]
for w in widgets:
for c in (w.dict().get("params") or {}).get("counters") or []:
print("widget title:", c.get("value"))
PY
}

BEFORE — origin/main: every counter says 'Drift in column 'value'', no group/label suffix.

git checkout origin/main -- src/evidently/metrics/column_statistics.py
run_check

Expected: 'widget title: Drift in column \'value\'' (twice).

AFTER — this branch: each widget shows the GroupBy decoration.

git checkout _pr1706 -- src/evidently/metrics/column_statistics.py
run_check

Expected: 'widget title: Value drift for value group by \'group\' for label: \'a\'' and same for label 'b'.

git checkout origin/main -- src/evidently/metrics/column_statistics.py
```

What I ran locally

• `pytest tests/future/metrics tests/test_preset/test_data_drift_preset.py -q` -> 1417 passed.
• `pytest tests/future/metrics/test_groupby_value_drift.py` -> 2 passed; on
  `origin/main` the GroupBy assertion fails with the bare "Drift in column 'value'"
  title.
• `ruff check` and `ruff format --check` -> clean.

Edge cases

Scenario	Before	After
Plain `ValueDrift(column="x")`	"Drift in column 'x'"	unchanged
`GroupBy(ValueDrift(column="x"), "g")`	"Drift in column 'x'" (bug)	"Value drift for x group by 'g' for label: '<label>'"
`GroupBy(MaxValue("x"), "g")` (other metric)	already correct	unchanged

AI disclosure

This pull request was authored with assistance from Anthropic's Claude (an AI
coding assistant) running under my direction. I traced the patching mechanism
(`metrics/group_by.py:_patched_display_name`) and confirmed the fix matches
the convention StatisticsCalculation already uses.