add size management and automatic pid naming

scripts/f_grad.sh

 source blaar.sh
 
+echo "f_grad.sh"
+echo "========="
+
 blaar_init_options "Apply a gradient on input grey image"
 blaar_add_option direction d "" "x|y" "direction of the gradient"
 blaar_add_option gain g "" "real" "gain" "1"
-blaar_add_option output o "" "blc_channel-out" "convolution output" ":grad<pid>"
+blaar_add_option output o "" "blc_channel-out" "convolution output" ":grad$$"
 blaar_add_option size s "" "(integer)x(interger)" "size of the filter" "3x3"
 blaar_add_option sigma S "" "real" "sigma of the filter" -1
 blaar_add_option profile p "" "filename" "profile the script" ""
@@ -12,14 +15,16 @@ blaar_parse_args "$@"
 
 [ "$direction" = "x" ] && option="-x1" || { [ "$direction" = "y" ] && option="-y1"; } || { echo "Wrong direction option '$direction'"; exit 1;  }
 
+echo "size:$size" >&2
 echo "Profile:$profile" >&2
 [ "$profile" != "" ] && profile_option="-~$profile" || profile_option=""
 
+
 kernel=$(i_gaussian $option -g$gain -s$size  -S$sigma -o/d${direction}gaussian$$ -p0 2>$TMP_DIR/log ) || { cat $TMP_DIR/log; echo "Error creating the kernel"; exit 2; }
 # -p0 is because we do not want block
 
 #./run.sh ipp/
-f_ipp_convolute $profile_option -o$output  $image $kernel
+f_ipp_convolute $profile_option -o$output $image $kernel
 
 #./run.sh armadillo/f_arma_convolute -o$output -~grad.tsv  $image $kernel
 #./run.sh bapps/f_conv -o$output -~grad.tsv  $image $kernel