add ambiguous claims to contradictions when penalized is true

[miafig]

Problem

When penalize_ambiguous_claims=True in FaithfulnessMetric, the reason generated is positive even if when there are penalizations from ambiguous claims.
This happens because the ambiguous claims are not counted as contradictions in _generate_reason().

Fix

Add all ambiguous claims to the contradictions list when penalize_ambiguous_claims=True.

from deepeval.metrics import FaithfulnessMetric
from deepeval.test_case import TestCase

example = LLMTestCase(
    intput="What is the refund policy?",
    actual_output="You can return items within 30 days.",
    retrieval_context=["Return policy 20 days for AMER", "Return policy 30 days for EMEA"],
    )

# Current behaviour
metric = FaithfulnessMetric(penalize_ambiguous_claims=True)
metric.measure(example)
>> 0.0
metric.reason
>> Awesome job! There are no contradictions ...

# New behaviour
metric = FaithfulnessMetric(penalize_ambiguous_claims=True)
metric.measure(example)
>> 0.0
metric.reason
>> The score is 0.00 because the actual output incorrectly claims ...

[vercel]

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

A member of the Team first needs to authorize it.

[penguine-ip]

@miafig thank you this is a nice addition, this will make it to the next release :)