Feat: Batch Streaming & Progressive Rendering for Proxy Loader

Objective

Optimize the Neo.data.proxy.Stream and Neo.data.Store integration to solve the performance regression caused by single-record event firing. Implement chunked streaming and progressive UI updates.

Problem

The initial implementation fired a data event for every single record in the NDJSON stream. For 11k records, this caused massive overhead (11k events, 11k store updates, 11k microtasks), blocking the App Worker for ~10s.

Solution

1. Batching (chunkSize): Update Stream.mjs to buffer parsed records and fire the data event only when chunkSize (e.g., 500) is reached.
2. Progressive Rendering: Update Store.mjs to remove suspendEvents during the stream. Instead, listen for the chunked data events, add the chunk to the store, and fire a load event immediately. This allows the Grid to render the first chunk (~500 rows) almost instantly while the rest of the stream continues in the background.

Tasks

• Add chunkSize config to Neo.data.proxy.Stream.
• Implement buffering logic in Stream.read() to yield arrays of records.
• Refactor Neo.data.Store.load() to support progressive loading (remove suspendEvents, fire intermediate load events).
• Update Unit Tests (Stream.spec.mjs, StoreProxy.spec.mjs).

Expected Outcome

• Time to First Byte (TTFB): Unchanged.
• Time to First Render (TTFR): Drastically reduced (O(chunkSize) instead of O(Total)).
• UX: User sees the grid populate immediately. Scrollbar grows as data streams in.