🐛 Unhandled JSON parsing in AI services and dataset processing

[morningstarxcdcode]

📋 Description\nMultiple AI service methods and dataset processing code call (or assume parsed JSON) without a robust fallback when the content is malformed or not JSON. This may cause runtime errors or unexpected behavior if external responses are not valid JSON.\n\n---\n\n## 🔄 Steps to Reproduce\n1. Mock an AI provider to return malformed JSON or a string that is not JSON.\n2. Call the endpoint that uses the AI service (e.g., ChartWithGroq / GenerateText endpoints).\n3. Observe unhandled parsing errors or exceptions.\n\n---\n\n## ✅ Expected Behavior\nThe code should safely attempt to parse JSON and fall back to returning the raw string or a safe default when parsing fails. It should also log diagnostics and avoid throwing unhandled parse exceptions.\n\n---\n\n## 🚫 Actual Behavior\n- Several places perform or assume is JSON and call directly.\n\nAffected files (initial scan):\n- \n- \n- \n- \n\n---\n\n## ✅ Fix / Recommendation\nIntroduce a single utility that returns a fallback value when parsing fails, log the error for diagnostics, and use it across the affected services. Add unit tests for the utility and update services to use it.\n\n---\n\n## 💻 Environment\n- Repo: main branch\n\n---\n\nI will open a PR from my fork implementing this fix with tests. If you need more details or would like to handle this differently, please let me know.

[morningstarxcdcode]

Linked PR: #49 — I added the utility and applied it to Google, Groq, Ollama, and DataSet services in this PR. The PR body now includes 'Fixes #42' so the issue will be closed when the PR is merged. Please review when you have a chance.