Super Lightweight Data Management

AI Prompt: Regenerate “Datalite” Interface with All Features Create a fully functional single-page HTML application called Datalite using only vanilla JavaScript, standard CSS, and HTML — no frameworks or libraries. The interface should support: Multi-file Upload & Tabs Allow uploading multiple CSV or JSON files Display each dataset as a tab in the left sidebar, sorted alphabetically Switching tabs shows the corresponding editable table Editable Table View (per dataset) Render a full HTML table with inline contentEditable cells Include a filter row at the top using contentEditable table headers Always show a sticky blank row at the bottom that adds new entries dynamically Changes persist in memory and to localStorage Relationship Detection (Heuristics) Detect one-to-many and many-to-one relationships heuristically by matching: {tableName}_id → id Shared key values across tables Store relationships and persist them DBML-style Diagram View Render a schema diagram using floating draggable HTML blocks on an infinite canvas Each block shows the table name and its fields Allow the user to toggle the diagram view on/off Report Builder Panel (on right) Show a live business-style report when any table row is clicked Include the selected record and all related child records in clean, sectioned formatting Add a toggle to switch between the styled view and raw JSON Include child tables (and grandchildren, etc.) using relationship chains State Persistence Save all uploaded files, relationships, and edits to localStorage Auto-load on refresh Resizable Panels Allow the user to drag and resize: Sidebar (tabs) Table view (center) Report panel (right) Export Functions Add an "Export CSV" button for the currently active tab Add a "Clear Storage" button to reset the app Requirements:

No libraries (no Tailwind, Bootstrap, React, etc.) Single relatix.html file Fully working out-of-the-box Mobile and desktop friendly Style should be minimal and clean (standard CSS only)