Support .txt files in LLMFileAnalysisOperator - #70431

Summary

LLMFileAnalysisOperator rejects .txt files with LLMFileAnalysisUnsupportedFormatError, but a file with no extension
at all (e.g. app) is already accepted and treated as plain text via the "log" fallback in detect_file_format(). The more explicitly-named file was the one that failed.

This also put LLMFileAnalysisOperator out of step with the rest of the provider:

• DocumentLoaderOperator (this provider's other file-reading operator) already lists .txt as one of its built-in, zero-extra-dependency formats (EXTENSION_BACKEND_MAP: ".txt": "text").
• Several of the provider's own example DAGs (example_llamaindex_rag.py, example_llamaindex_hook.py, example_langchain_tool_agent.py) ship .txt files as sample input for LLM-related tasks.

Changes

• Add "txt" to SUPPORTED_FILE_FORMATS, _TEXT_LIKE_FORMATS, and _GZIP_SUPPORTED_FORMATS in utils/file_analysis.py.
• No parsing/rendering logic changes: the existing generic text-renderingpath (_render_text_like, already used for .log and extension-less files) handles .txt as-is.