Unify PDF and image conversion engines

[PrzemyslawKlys]

Summary

• route PDF image placement through one shared raster preparation path so authored/stamped PDFs and Word, Excel, Markdown, and RTF adapters consistently accept JPEG, PNG, GIF, BMP, TIFF, and supported WebP payloads without duplicate normalization
• add lazy, cancellation-safe five-format image export directly from authored PDF documents and a typed component contract that reuses the existing flow/layout engine
• add a direct loss-aware OpenDocument-to-PDF package for text, spreadsheet, and presentation documents, preserving source-projection and PDF-stage diagnostics
• generate a public conversion support matrix from the scenario manifest, require implementation/test evidence for direct routes, and document current fidelity plus the remaining engine work in easiest-to-most-complex order
• harden the shared atomic image-export commit path against transient Windows sharing violations so create-unique saves remain deterministic under concurrent writers

Validation

• dotnet build OfficeIMO.sln -c Release --no-restore — 0 warnings, 0 errors
• PDF tests — 3,068 passed on .NET 8 and 3,068 on .NET 10
• Drawing tests — 916 passed on .NET 8 and 916 on .NET 10
• concurrent image-export contracts — 25 repeated passes on .NET 8 and .NET 10; .NET Framework 4.7.2 compile passed with 0 warnings and 0 errors
• RTF tests — 696 passed on .NET 8 and 696 on .NET 10
• OpenDocument converter tests — 45 passed on .NET 8 and 45 on .NET 10
• release packaging guardrails — 5 passed on each target
• generated support-matrix drift check passed
• OfficeIMO.OpenDocument.Pdf packed successfully with .NET Standard 2.0, .NET 8, and .NET 10 assets
• independent read-only review completed, followed by one targeted confirmation of all accepted fixes with no unresolved actionable findings