IconTK: Simplify icon clipping logic more. Saves 15 bytes.

Same as previous commit, but for the window bounds right edge.

In the preamble to the complicated icon clipping logic, the right
bounds of the overlapping window is decremented by a pixel before
diving in to determine how to adjust the clipping rect. After staring
at the math for a bit, it turns out that it is because of how "greater
than" test is done - the test was implemented as a "greater than or
equals" (c/o BMI, skipping if "less than"). If the test is inverted
then it can properly be a "greater than" (c/o BPL, skipping if
"greater than or equal") and the decrement can be skipped and a copy
further simplified.

Then there's a follow-on adjustment needed as well, similar but
inverted test, and a +2 becomes a +1 which makes more sense. Yay!

toolkits/icontk.s

@@ -2223,9 +2223,6 @@ do_pt:  lda     pt_num
         copy    findwindow_params::window_id, getwinframerect_params::window_id
         MGTK_CALL MGTK::GetWinFrameRect, getwinframerect_params
 
-        ;; TODO: Determine why these are necessary:
-        dec16   win_r
-
         ;; ==================================================
         ;; At this point, win_r/t/l/b are the window edges,
         ;; cr_r/t/l/b are the rect we know has at least one
@@ -2252,17 +2249,11 @@ do_pt:  lda     pt_num
 
         ;; Cases 1/2/3 (and continue below)
         ;; if (cr_r > win_r)
-        ;; . cr_r = win_r + 1
-        scmp16  cr_r, win_r
-        bmi     case789
+        ;; . cr_r = win_r
+        scmp16  win_r, cr_r
+        bpl     case789
 
-        ldx     win_r
-        ldy     win_r + 1
-        inx
-        bne     :+
-        iny
-:       stx     cr_r
-        sty     cr_r + 1
+        copy16  win_r, cr_r
         jmp     vert
 
         ;; Cases 7/8/9 (and done)
@@ -2306,16 +2297,16 @@ vert:   scmp16  cr_t, win_t
 
         ;; Case 2
         ;; if (win_r < stash_r)
-        ;; . cr_l = win_r + 2
-        ;; . vx   = win_r + 2
+        ;; . cr_l = win_r + 1
+        ;; . vx   = win_r + 1
         ;; . cr_r = stash_r + 2 (workaround for https://github.com/a2stuff/a2d/issues/153)
 case2:
-        scmp16  win_r, stash_r
-        bpl     :+
+        scmp16  stash_r, win_r
+        bmi     :+
 
         lda     win_r
         clc
-        adc     #2
+        adc     #1
         sta     cr_l
         sta     vx
         lda     win_r+1