dda.c: no cruising time when rampup_steps == move_step_no

In some cases when move_step_no is equal to rampup_steps, this
algorithm think it is cruising. When cruising it setup the dda->c
to dda->c_min. Which is wrong in that case.

So we let recalculate the dda->c now. The axis will become a little
bit faster for none-cruising movements. When it hits cruising,
it will be capped anyway to dda->c_min. So the "TODO: check is obsolete"
is not obsolete anymore.

dda.c

@@ -865,7 +865,7 @@ void dda_clock() {
     move_step_no = dda->total_steps - move_step;
 
     recalc_speed = 0;
-    if (move_step_no < dda->rampup_steps) {
+    if (move_step_no <= dda->rampup_steps) {
       #ifdef LOOKAHEAD
         move_n = dda->start_steps + move_step_no;
       #else
@@ -896,9 +896,6 @@ void dda_clock() {
                   int_inv_sqrt(move_n)) >> 13;
         #endif
 
-      // TODO: most likely this whole check is obsolete. It was left as a
-      //       safety margin, only. Rampup steps calculation should be accurate
-      //       now and give the requested target speed within a few percent.
       if (move_c < dda->c_min) {
         // We hit max speed not always exactly.
         move_c = dda->c_min;