Guard some truncating casts with & 0xFF

The result of adding two unsigned chars is an int, so assigning this again
to unsigned char results in a narrowing cast. Afaik this is well-defined
by the standard but code compiled with cl with all runtime checks enabled
will trigger a breakpoint when e.g. scanline[i] + recon[i - bytewidth]
yields a number > 255. Fix this, and make it clear what's happening, by
and'ing with 0xFF (which should be discarded by any sane optimizer anyway).

lodepng.cpp

@@ -3980,12 +3980,12 @@ static unsigned unfilterScanline(unsigned char* recon, const unsigned char* scan
       break;
     case 1:
       for(i = 0; i != bytewidth; ++i) recon[i] = scanline[i];
-      for(i = bytewidth; i < length; ++i) recon[i] = scanline[i] + recon[i - bytewidth];
+      for(i = bytewidth; i < length; ++i) recon[i] = (scanline[i] + recon[i - bytewidth]) & 0xFF;
       break;
     case 2:
       if(precon)
       {
-        for(i = 0; i != length; ++i) recon[i] = scanline[i] + precon[i];
+        for(i = 0; i != length; ++i) recon[i] = (scanline[i] + precon[i]) & 0xFF;
       }
       else
       {
@@ -3995,25 +3995,25 @@ static unsigned unfilterScanline(unsigned char* recon, const unsigned char* scan
     case 3:
       if(precon)
       {
-        for(i = 0; i != bytewidth; ++i) recon[i] = scanline[i] + (precon[i] >> 1);
-        for(i = bytewidth; i < length; ++i) recon[i] = scanline[i] + ((recon[i - bytewidth] + precon[i]) >> 1);
+        for(i = 0; i != bytewidth; ++i) recon[i] = (scanline[i] + (precon[i] >> 1)) & 0xFF;
+        for(i = bytewidth; i < length; ++i) recon[i] = (scanline[i] + ((recon[i - bytewidth] + precon[i]) >> 1)) & 0xFF;
       }
       else
       {
         for(i = 0; i != bytewidth; ++i) recon[i] = scanline[i];
-        for(i = bytewidth; i < length; ++i) recon[i] = scanline[i] + (recon[i - bytewidth] >> 1);
+        for(i = bytewidth; i < length; ++i) recon[i] = (scanline[i] + (recon[i - bytewidth] >> 1)) & 0xFF;
       }
       break;
     case 4:
       if(precon)
       {
         for(i = 0; i != bytewidth; ++i)
         {
-          recon[i] = (scanline[i] + precon[i]); /*paethPredictor(0, precon[i], 0) is always precon[i]*/
+          recon[i] = (scanline[i] + precon[i]) & 0xFF; /*paethPredictor(0, precon[i], 0) is always precon[i]*/
         }
         for(i = bytewidth; i < length; ++i)
         {
-          recon[i] = (scanline[i] + paethPredictor(recon[i - bytewidth], precon[i], precon[i - bytewidth]));
+          recon[i] = (scanline[i] + paethPredictor(recon[i - bytewidth], precon[i], precon[i - bytewidth])) & 0xFF;
         }
       }
       else
@@ -4025,7 +4025,7 @@ static unsigned unfilterScanline(unsigned char* recon, const unsigned char* scan
         for(i = bytewidth; i < length; ++i)
         {
           /*paethPredictor(recon[i - bytewidth], 0, 0) is always recon[i - bytewidth]*/
-          recon[i] = (scanline[i] + recon[i - bytewidth]);
+          recon[i] = (scanline[i] + recon[i - bytewidth]) & 0xFF;
         }
       }
       break;