Add retry mechanism for Quix Portal API connection errors

[gwaramadze]

Problem

The Quix Portal API calls were failing with DNS resolution errors (httpx.ConnectError) during temporary network issues, causing applications to crash instead of gracefully handling transient connectivity problems.

Solution

Implemented a retry mechanism with exponential backoff for connection-related errors:

• Retry decorator with 5 attempts and exponential backoff (2s, 4s, 8s, 16s, 32s)
• Applied to all 8 HTTP methods in QuixPortalApiService
• Handles both DNS errors and timeouts while preserving fast-fail for legitimate API errors (404, 401, etc.)
• Comprehensive logging for retry attempts and failures

Changes

• Added retry_on_connection_error decorator in quixstreams/platforms/quix/api.py
• Applied decorator to all Portal API methods (get_librdkafka_connection_config, get_workspace, etc.)
• Added 4 tests covering success scenarios, max retries, and different error types

This resolves intermittent DNS and connection timeout issues when connecting to Quix Cloud, making the SDK more resilient to network instability.