fix: preserve alias in metric tests for all generic test helpers

[CTWalk]

Thanks for maintaining Evidently. This is a small consistency fix in the
generic test helpers; the reproduction and reasoning are below.

Related issue: #1907

What

gte() forwards alias= into its metric test. The seven sibling helpers
(eq, not_eq, gt, lt, lte, is_in, not_in) also accept alias=,
but forward it only to DescriptorTest. When one of them is used as a
metric-level test in a Report, the configured alias is silently dropped.

Why it matters

MetricTest.alias is the custom name used to build the test description:

# src/evidently/core/metric_types.py
description = f"{self.alias or value.display_name}: {result.description}"

For seven of the eight helpers that field can never be set through the public
helper, so a name the user configured never reaches the report.

Reproduce

On main (a4aa4c2, 0.7.21):

import pandas as pd

from evidently import Report
from evidently.metrics import MeanValue
from evidently.tests import gt, gte

data = pd.DataFrame({"score": [0.0, 1.0]})
for helper in (gte, gt):
    test = helper(1, alias="quality gate")
    snapshot = Report([MeanValue(column="score", tests=[test])]).run(data)
    print(helper.__name__, test.metric.alias)
    print(" ", snapshot.tests_results[0].description)

Before:

gte quality gate
  quality gate: Actual value 0.500 < 1.000
gt None
  Mean value of 'score': Actual value 0.500 < 1.000

After:

gte quality gate
  quality gate: Actual value 0.500 < 1.000
gt quality gate
  quality gate: Actual value 0.500 < 1.000

Change

• src/evidently/tests/aliases.py — pass the already accepted alias
  argument into the metric-test constructor in the seven helpers. gte is
  untouched and serves as the reference behavior. ruff format rewraps five
  of those calls onto multiple lines; the only semantic change is
  alias=alias.
• tests/future/tests/test_generic_tests_creation.py — one parameterized
  regression over all eight helpers, asserting the alias survives on both the
  metric and the descriptor.

No new API, parameter, state, or dependency. alias defaults to None, so
behavior is unchanged for every caller that does not pass it.

Checks

Run locally against this branch on Python 3.12:

• pytest tests/future/tests/test_generic_tests_creation.py — 7 failed, 8 passed before the source change, 15 passed after. Reverting only the
  source change brings the seven failures back, so the new test does pin the
  behavior.
• All non-UI tests/future — 1626 passed, 6 skipped, no failures.
• ruff check and ruff format --check on both files, using the repository's
  ruff.toml and pinned ruff==0.3.7.

Every other alias= usage in the test tree is descriptor-level, so no
existing test exercises the changed metric path.

Not run locally: tests/future/test_ui (needs sqlalchemy), the legacy
tests/ tree, and mypy.

One open question

The @overload signatures in evidently/tests/aliases.py currently declare
alias only on the DescriptorTest-returning forms, so a typed
gt(1, alias="quality gate") resolves to the descriptor overload. This PR
deliberately leaves the overloads alone — the same gap already applies to
gte(), so nothing gets worse — but if you would like the metric-shaped
overloads to declare alias as well, I am happy to extend this PR.