[ISSUE] LLM-NLU Helper: Language Entity Value Incorrectly Returns User Input

[medchedli]

When processing user input via the /predict endpoint of the LLM NLU helper, there are instances where the system correctly identifies the presence of a language entity but fails to populate its value field with the classified language code (e.g., 'en', 'fr'). Instead, the value field incorrectly contains the literal user input string that was being analyzed.