Improve chart engine, insight engine, LLM client, question planner and JSON utils

fix(backend): parse cors_origins from JSON or comma-separated string

feat: update main application entrypoint

feat: enhance markdown report engine and exports

feat: integrate sql planner into questions router

feat: update dataset router to support custom metadata

feat: add sql planning capabilities to question planner

feat: refine llm client and deepseek api calls

feat: extend analysis plan executor with sql support

feat: add sql planner using deepseek reasoner

feat: implement sql execution engine with duckdb

feat: add robust json parsing utilities

feat: implement centralized backend config options

chore: configure local environment settings

chore: add duckdb dependency to requirements.txt

docs: update readme with huggingface space configuration

feat: add dockerfile and dockerignore for backend deployment

chore: remove frontend components and utilities

chore: remove frontend page routes

chore: remove frontend global styles and base layout

chore: remove frontend build and transpiler configs

chore: remove frontend package configuration

docs: add final project documentation

test: add sample data and demo workflow

chore: add validation and safety guardrails

style: polish frontend product experience

feat: add report generation system

feat: add autonomous insights engine

feat: add chart system

feat: add result explanation engine

feat: execute analysis plans with pandas numpy

feat: add LLM question planner

feat: add pandas numpy analysis engine

feat: add dataframe store

feat: add dataset profiling system

feat: add dataset upload system

chore: initialize InsightGrid AI project