refactor(infrastructure): implement Stream-based PDF processing

DevExpressPdfProcessor:
- ValidateAsync, ValidatePdfAAsync, CheckAttachmentsAsync: Stream parameters
- Defensive Position=0 validation (BadRequestException for seekable streams not at beginning)
- Remove unsafe Position reset (non-seekable stream compatibility)
- Remove PdfProcessingException wrapping (let DevExpress exceptions propagate naturally)

DevExpressSwissQrCodeProcessor:
- ExtractSwissQrCodeAsync: Stream parameter
- Defensive Position=0 validation
- Remove unsafe Position reset

Memory optimization: MemoryStream.TryGetBuffer fast path for byte[] extraction

DocumentOperator.Infrastructure/Services/PdfProcessing/DevExpressPdfProcessor.cs

@@ -28,6 +28,12 @@ public class DevExpressPdfProcessor : IPdfProcessor
             throw new BadRequestException("PDF stream cannot be empty");
         }
 
+        // Defensive validation: Seekable streams must be at Position = 0
+        if (pdfStream.CanSeek && pdfStream.Position != 0)
+        {
+            throw new BadRequestException("PDF stream must be positioned at the beginning (Position = 0).");
+        }
+
         // 2. Read stream to byte array for raw data analysis
         // (DevExpress needs byte[] for some operations like attachment detection)
         byte[] pdfBytes;
@@ -39,14 +45,18 @@ public class DevExpressPdfProcessor : IPdfProcessor
         else
         {
             // Slow path: copy stream to byte array
-            pdfStream.Position = 0;
             using var memoryStream = new MemoryStream();
             await pdfStream.CopyToAsync(memoryStream);
             pdfBytes = memoryStream.ToArray();
         }
 
         // 3. Load PDF with DevExpress Document API
-        pdfStream.Position = 0;
+        // Reset position for DevExpress (seekable streams only)
+        if (pdfStream.CanSeek)
+        {
+            pdfStream.Position = 0;
+        }
+        
         using var processor = new PdfDocumentProcessor();
         processor.LoadDocument(pdfStream);
 
@@ -87,6 +97,12 @@ public class DevExpressPdfProcessor : IPdfProcessor
             throw new BadRequestException("PDF stream cannot be empty");
         }
 
+        // Defensive validation: Seekable streams must be at Position = 0
+        if (pdfStream.CanSeek && pdfStream.Position != 0)
+        {
+            throw new BadRequestException("PDF stream must be positioned at the beginning (Position = 0).");
+        }
+
         // 2. Read stream to byte array for raw data analysis
         byte[] pdfBytes;
         if (pdfStream is MemoryStream ms && ms.TryGetBuffer(out var buffer))
@@ -95,14 +111,18 @@ public class DevExpressPdfProcessor : IPdfProcessor
         }
         else
         {
-            pdfStream.Position = 0;
             using var memoryStream = new MemoryStream();
             await pdfStream.CopyToAsync(memoryStream);
             pdfBytes = memoryStream.ToArray();
         }
 
         // 3. Load PDF with DevExpress Document API
-        pdfStream.Position = 0;
+        // Reset position for DevExpress (seekable streams only)
+        if (pdfStream.CanSeek)
+        {
+            pdfStream.Position = 0;
+        }
+        
         using var processor = new PdfDocumentProcessor();
         processor.LoadDocument(pdfStream);
 
@@ -175,8 +195,15 @@ public class DevExpressPdfProcessor : IPdfProcessor
             throw new BadRequestException("PDF stream cannot be empty");
         }
 
-        // 2. Load PDF with DevExpress Document API (exceptions propagate naturally)
-        pdfStream.Position = 0;
+        // Defensive validation: Seekable streams must be at Position = 0
+        if (pdfStream.CanSeek && pdfStream.Position != 0)
+        {
+            throw new BadRequestException("PDF stream must be positioned at the beginning (Position = 0).");
+        }
+
+        // 2. Load PDF with DevExpress Document API
+        // DevExpress LoadDocument may throw exceptions for corrupted PDFs - let them propagate naturally
+        // Middleware will catch and convert to 500 Internal Server Error
         using var processor = new PdfDocumentProcessor();
         processor.LoadDocument(pdfStream);
 

DocumentOperator.Infrastructure/Services/QrCodeProcessing/DevExpressSwissQrCodeProcessor.cs

@@ -24,19 +24,27 @@ public sealed class DevExpressSwissQrCodeProcessor : ISwissQrCodeProcessor
 
     /// <inheritdoc />
     public async Task<(Bill Bill, string[] RawLines)> ExtractSwissQrCodeAsync(
-        byte[] pdfBytes,
+        Stream pdfStream,
         int[]? pageNumbers = null,
         CancellationToken cancellationToken = default)
     {
-        if (pdfBytes.Length == 0)
-            throw new ArgumentException("PDF document contains no byte data.", nameof(pdfBytes));
+        ArgumentNullException.ThrowIfNull(pdfStream, nameof(pdfStream));
+        
+        if (pdfStream.Length == 0)
+            throw new ArgumentException("PDF stream is empty.", nameof(pdfStream));
+
+        // Defensive validation: Seekable streams must be at Position = 0
+        // Non-seekable streams (e.g., NetworkStream) are not checked
+        if (pdfStream.CanSeek && pdfStream.Position != 0)
+            throw new BadRequestException(null, new ArgumentException(
+                "PDF stream must be positioned at the beginning (Position = 0).",
+                nameof(pdfStream)));
 
         using var pdfDocument = new PdfDocumentProcessor();
-        using var pdfStream = new MemoryStream(pdfBytes);
         pdfDocument.LoadDocument(pdfStream);
 
         if (pdfDocument.Document.Pages.Count == 0)
-            throw new ArgumentException("PDF document contains no pages.", nameof(pdfBytes));
+            throw new ArgumentException("PDF document contains no pages.", nameof(pdfStream));
 
         // Determine which pages to scan
         int[] pagesToScan = DeterminePageNumbers(pdfDocument.Document.Pages.Count, pageNumbers);