fix: prevent memory leaks in streaming chat responses

[Vijaykv5]

Fixes : #466

What this PR does :

Added buffer size limits and memory management to prevent unbounded growth during long-running streaming sessions.
It Implements1MB buffer limit, 16KB chunk truncation, and 10MB total message size capacity to maintain browser performance during code analysis.

[Copilot]

Pull Request Overview

This PR addresses memory leak issues in streaming chat responses by implementing buffer size limits and memory management controls. The changes prevent unbounded memory growth during long-running streaming sessions by enforcing strict limits on buffer sizes and message processing.

• Adds configurable memory limits (1MB buffer, 16KB chunk truncation, 10MB total message size)
• Implements buffer truncation logic to prevent excessive memory usage
• Adds error handling for messages exceeding size limits

[Copilot]

The processedSize is tracking the raw byte length of chunks, but buffer length is measured in characters after decoding. This mismatch could cause incorrect size calculations, especially with multi-byte UTF-8 characters. Consider tracking decoded character length instead or use value.byteLength for consistent byte-based measurement.

Suggested change

	processedSize += value.length;
	processedSize += value.byteLength;

[Copilot]

Using slice(-CHUNK_SIZE) could truncate in the middle of a multi-byte UTF-8 character, potentially corrupting the data. This could lead to decoding errors or malformed SSE events. Consider implementing UTF-8 safe truncation or truncating at character boundaries.

[Copilot]

Using console.warn for production logging may not be appropriate. Consider using a proper logging framework or callback mechanism to handle this warning, as it could indicate data loss that the application should handle gracefully.