Support PDF in LLMTestCase input for agentic evaluation

[anuar12]

Problem

LLMTestCase.input only accepts str, making it impossible to include PDF documents as part of the agent input.
We are using pdf input like here: https://developers.openai.com/api/docs/guides/file-inputs/

Proposed Solution

Extend multimodal support to PDFs, following the existing MLLMImage pattern:

from deepeval.test_case import LLMTestCase, MLLMDocument

pdf = MLLMDocument(url='./contract.pdf', local=True)

test_case = LLMTestCase(
    input=f"What does this contract say about payment terms? {pdf}",
    tools_called=[...]
)