StreamThreadException investigation and fix (spike, do not merge)

Summary

This spike PR investigates and proposes a fix for issue #8301 - a StreamThreadException in the Bing Ads source connector where the campaign_labels stream fails with:

'utf-8' codec can't decode byte 0x8b in position 1: invalid start byte

Root Cause: Byte 0x8b is the GZIP magic number, indicating that GZIP-compressed data is being incorrectly passed to a UTF-8 decoder. This occurs when the CompositeRawDecoder's parser selection logic fails to detect GZIP content (likely due to missing Content-Encoding headers), causing compressed data to be treated as plain text.

Proposed Solution: Enhanced CompositeRawDecoder with auto-detection of GZIP content by magic bytes, better error handling, and graceful fallback mechanisms.

Review & Testing Checklist for Human

🔴 High Risk - 5 Critical Items

• Validate Root Cause Analysis: Review the investigation document and confirm that GZIP magic byte detection is the correct approach for this issue
• Test Reproduction Script: Run test_gzip_utf8_issue.py in a proper environment to verify it reproduces the exact error (had import issues locally)
• Evaluate Integration Strategy: Decide whether to extend existing CompositeRawDecoder or create new implementation - current proposal creates separate class
• End-to-End Testing: Test the proposed fix with actual Bing Ads connector campaign_labels stream to verify it resolves the issue
• Impact Assessment: Review if auto-detection of GZIP content could affect other connectors that use CompositeRawDecoder

Recommended Test Plan:

1. Set up Bing Ads connector with valid credentials
2. Sync campaign_labels stream to reproduce the error
3. Apply the proposed fix and verify error is resolved
4. Test other Bing Ads streams to ensure no regressions
5. Run connector test suite to validate broader impact

Diagram

%%{ init : { "theme" : "default" }}%%
graph TD
    Issue["Issue #8301<br/>StreamThreadException<br/>campaign_labels"]
    
    BingAds["airbyte/airbyte-integrations/<br/>connectors/source-bing-ads/<br/>manifest.yaml"]:::context
    CompositeDecoder["airbyte_cdk/sources/declarative/<br/>decoders/composite_raw_decoder.py"]:::context
    ConcurrentSource["airbyte_cdk/sources/concurrent_source/<br/>concurrent_read_processor.py"]:::context
    
    Investigation["SPIKE_INVESTIGATION.md"]:::major-edit
    TestScript["test_gzip_utf8_issue.py"]:::major-edit
    ProposedFix["fix_gzip_parser_selection.py"]:::major-edit
    
    Issue --> Investigation
    Issue --> TestScript
    Issue --> ProposedFix
    
    BingAds --> CompositeDecoder
    CompositeDecoder --> ConcurrentSource
    ConcurrentSource --> Issue
    
    Investigation --> CompositeDecoder
    TestScript --> CompositeDecoder
    ProposedFix --> CompositeDecoder
    
    subgraph Legend
        L1[Major Edit]:::major-edit
        L2[Minor Edit]:::minor-edit
        L3[Context/No Edit]:::context
    end
    
    classDef major-edit fill:#90EE90
    classDef minor-edit fill:#87CEEB
    classDef context fill:#FFFFFF

Notes

• Session Details: Requested by @aaronsteers | Devin Session
• Spike Nature: This is investigation work with a proposed solution - not production-ready code
• Import Issues: Test script had dependency issues when run locally, may need environment setup
• Architecture Decision: Proposed fix creates new class rather than modifying existing CompositeRawDecoder - integration approach needs decision
• Limited Testing: No actual Bing Ads connector testing performed - purely theoretical based on code analysis
• Concurrent Framework: Issue occurs specifically in concurrent source processing, not traditional declarative streams