refactor: Avoid undefined behavior (signed int overflow) in negations

The computation of `abs(x)` (where the type of `x` is `png_int_32`)
should involve a cast to `png_uint_32` before negating `x`, ensuring
that the result is well-defined even for `INT32_MIN`.

Considering that the PNG Specification (and, implicitly, libpng)
formally prohibits `INT32_MIN` values, this is merely a zero-cost form
of hardening intended to appease UBSan, and not an actual bug fix.

pngread.c

@@ -4111,7 +4111,7 @@ png_image_finish_read(png_imagep image, png_const_colorp background,
             row_stride = (png_int_32)/*SAFE*/png_row_stride;
 
          if (row_stride < 0)
-            check = (png_uint_32)(-row_stride);
+            check = -(png_uint_32)row_stride;
 
          else
             check = (png_uint_32)row_stride;

pngwrite.c

@@ -2030,7 +2030,7 @@ png_image_write_main(png_voidp argument)
             display->row_stride = (png_int_32)/*SAFE*/png_row_stride;
 
          if (display->row_stride < 0)
-            check = (png_uint_32)(-display->row_stride);
+            check = -(png_uint_32)display->row_stride;
 
          else
             check = (png_uint_32)display->row_stride;