add documentation for issue 4148

Per Claude:

The key things the new comments add that the old ones were missing:

1. Palette RGB colors are completely ignored — only the index value is used as alpha. This is the non-obvious invariant that @notimaginative described.
2. Exact mapping table — makes the * 18 / * 17 math immediately readable, and clarifies the ANI vs PCX difference (> 15 clamps to opaque in ANI but transparent in PCX).
3. The broken-palette consequence — explicitly states that a reversed palette causes a white square, linking to #4148 so future readers understand the context.

code/anim/packunpack.cpp

@@ -142,11 +142,15 @@ int unpack_pixel(anim_instance *ai, ubyte *data, ubyte pix, int aabitmap, int bp
 			bit_16 = (ushort)pix;
 			break;
 		case 8:
-			// 8 bit-per-pixel aa bitmaps are a bit special since they only use a palette index value in the range [0, 15]. These 
-			// palette indexes must be remapped to alpha values between [0, 255] which is what graphics code expects. Palette 
-			// range [0, 14] is a gradient from black to white, and palette index 15 is a special color which indicates the background
-			// area of a HUD gauge. Retail code uses the final alpha value for index 1 for this special index to give gauges a dark
-			// transparent background.
+			// 8-bit-per-pixel HUD gauge ANI aabitmaps use the palette index directly as an alpha value —
+			// the actual RGB colors stored in the palette are completely ignored. The valid range is [0, 15]:
+			//   [0]:    fully transparent (alpha 0)
+			//   [1-14]: black-to-white gradient (alpha = index * 18)
+			//   [15]:   special HUD background (alpha 18, matching index 1, for a dark transparent tint)
+			//   [>15]:  clamped to fully opaque (alpha 255)
+			// Because only the index value matters, the palette in the source file must be ordered so that
+			// index 0 is the transparent end and index 14 is the opaque end. A reversed palette causes the
+			// gauge to render as a solid white square (see GitHub issue #4148).
 			if (pix > 15) {
 				bit_8 = 255;
 			}
@@ -240,11 +244,15 @@ int unpack_pixel_count(anim_instance *ai, ubyte *data, ubyte pix, int count = 0,
 			bit_16 = (ushort)pix;
 			break;
 		case 8 :
-			// 8 bit-per-pixel aa bitmaps are a bit special since they only use a palette index value in the range [0, 15]. These 
-			// palette indexes must be remapped to alpha values between [0, 255] which is what graphics code expects. Palette 
-			// range [0, 14] is a gradient from black to white, and palette index 15 is a special color which indicates the background
-			// area of a HUD gauge. Retail code uses the final alpha value for index 1 for this special index to give gauges a dark
-			// transparent background.
+			// 8-bit-per-pixel HUD gauge ANI aabitmaps use the palette index directly as an alpha value —
+			// the actual RGB colors stored in the palette are completely ignored. The valid range is [0, 15]:
+			//   [0]:    fully transparent (alpha 0)
+			//   [1-14]: black-to-white gradient (alpha = index * 18)
+			//   [15]:   special HUD background (alpha 18, matching index 1, for a dark transparent tint)
+			//   [>15]:  clamped to fully opaque (alpha 255)
+			// Because only the index value matters, the palette in the source file must be ordered so that
+			// index 0 is the transparent end and index 14 is the opaque end. A reversed palette causes the
+			// gauge to render as a solid white square (see GitHub issue #4148).
 			if (pix > 15) {
 				bit_8 = 255;
 			}

code/pcxutils/pcxutils.cpp

@@ -312,11 +312,16 @@ int pcx_read_bitmap( const char * real_filename, ubyte *org_data, ubyte * /*pal*
 				if ( byte_size == 1 ) {
 					auto pixel_val = data;
 					if (!mask_bitmap) {
-						// 8 bit-per-pixel aa bitmaps are a bit special since they only use values in the range [0, 15] where 15 wraps
-						// around back to 0. Since the rest of the code expects the value to be in the range [0, 255] the pixel value
-						// needs to be adjusted here. By multiplying the value with 17 the original range [0, 15] is mapped to [0, 255]
-						// This only applies to bitmaps that are not used as masks since mask bitmaps use higher values
-						// to indicate their mask area
+						// 8-bit-per-pixel HUD gauge PCX aabitmaps use the pixel value directly as an alpha level —
+						// the actual RGB colors in the palette are completely ignored. The valid range is [0, 15]:
+						//   [0]:    fully transparent (alpha 0)
+						//   [1-14]: black-to-white gradient (alpha = value * 17)
+						//   [15]:   special HUD background (alpha 17, matching index 1, for a dark transparent tint)
+						//   [>15]:  clamped to fully transparent (alpha 0)
+						// Because only the pixel value matters, the palette in the source file must be ordered so that
+						// index 0 is the transparent end and index 14 is the opaque end. A reversed palette causes the
+						// gauge to render as a solid white square (see GitHub issue #4148).
+						// Mask bitmaps skip this block because they use the full value range for the mask area.
 						if (data > 15) {
 							pixel_val = 0;
 						} else if (data == 15) {