fix: return 1.0 when no knowledge retention verdicts exist

[NgDMau]

Problem

KnowledgeRetentionMetric._calculate_score() returns 0 when self.verdicts
is empty. This happens in short conversations (1-2 turns) where there is no
accumulated knowledge to forget.

Zero verdicts means "nothing was forgotten" — the reason text confirms this
("no attritions... all knowledge retained") but the score contradicts it by
returning 0.

Fix

Return 1.0 instead of 0 when there are no verdicts. "Nothing to forget" is
a perfect retention score.

How to reproduce

from deepeval.metrics import KnowledgeRetentionMetric
from deepeval.test_case import ConversationalTestCase
from deepeval.test_case.conversational_test_case import Turn

test_case = ConversationalTestCase(
    turns=[
        Turn(role="user", content="What is WOVN?"),
        Turn(role="assistant", content="WOVN is a localization platform."),
    ]
)

metric = KnowledgeRetentionMetric()
metric.measure(test_case)
print(metric.score)   # 0.0 — should be 1.0
print(metric.reason)  # "no attritions" — contradicts the score

[vercel]

@NgDMau is attempting to deploy a commit to the Confident AI Team on Vercel.

A member of the Team first needs to authorize it.