Attempt at avoiding the false changes seen on Xue. Seems that alpha-blending

produces minute differences where there should be none.

- scripts/ppmdiff/ppmdiff.c (diff): instead of looking for all or noting
  changes, also detect gradual changes
- scripts/ppmdiff/ppmdiff.c (diff, main): don't mark areas with differences
  in the main images if using shadow images
- scripts/schps2ppm: new option to disable alpha blending
- scripts/schhist2web: cache pixmaps with default line thickness and no
  alpha blending in hard_*
- scripts/schhist2web: base marking of areas with differences exclusively
  on the hard_* pixmaps

scripts/schps2ppm

@@ -20,14 +20,18 @@ usage()
     cat <<EOF 1>&2
 usage: $0 [options] [file.ps [file.ppm]]
 
-  -r XxY     image resolution (default: $RES)
+  -n      disable alpha blending
+  -r XxY  image resolution (default: $RES)
 EOF
     exit 1
 }
 
 
+alpha=true
 while true; do
     case "$1" in
+    -n)	alpha=false
+	shift;;
     -r)	[ -z "$2" ] && usage
 	RES="$2"
 	shift 2;;
@@ -48,5 +52,6 @@ res=`expr 72 \* $X / 800`
 
 cat "$in" |
   gs -sDEVICE=ppmraw -sOutputFile=- -g$IRES -r$res \
-  -dTextAlphaBits=4 -dGraphicsAlphaBits=4 -q - |
+   `$alpha && echo '' -dTextAlphaBits=4 -dGraphicsAlphaBits=4` \
+   -q - |
   pnmflip -r270 >"$out"