community: Add cost per 1K tokens for fine-tuned model cached input (#…

…29248)

### Description

- Since there is no cost per 1k input tokens for a fine-tuned cached
version of `gpt-4o-mini-2024-07-18` is not available when using the
`OpenAICallbackHandler`, it raises an error when trying to make calls
with such model.
- To add the price in the `MODEL_COST_PER_1K_TOKENS` dictionary

cc. @efriis

libs/community/langchain_community/callbacks/openai_info.py

@@ -128,6 +128,7 @@
     "gpt-3.5-turbo-1106-finetuned": 0.003,
     "gpt-3.5-turbo-0125-finetuned": 0.003,
     "gpt-4o-mini-2024-07-18-finetuned": 0.0003,
+    "gpt-4o-mini-2024-07-18-finetuned-cached": 0.00015,
     # Fine Tuned output
     "babbage-002-finetuned-completion": 0.0016,
     "davinci-002-finetuned-completion": 0.012,