Handle floating-point numbers (closes holman#28)

Mostly I just want make this last test green. Shelling out to awk slows things
down, unfortunately, but at least at this point forward we can just improve the
performance. I still suspect we can simplify the whole thing with a smarter
rewrite and speed things up to boot.

spark

@@ -69,7 +69,7 @@ setup_array() {
 sort_max()
 {
   last=${#sorted[@]}
-  echo ${sorted[$last - 1]}
+  echo $(echo ${sorted[$last - 1]} | awk '{printf "%.0f",$1}')
 }
 
 # The minimum value of the sorted array. In other words, the first value.
@@ -100,7 +100,7 @@ print_tick()
   for (( i = 0 ; i < ${#ticks[@]} ; i++ ))
   do
 
-    number=$1
+    number=$(echo $1 | awk '{printf "%.0f",$1}')
     less_than=$(( $i * $tier + sort_min + $tier ))
     greater_than=$(( ($i - 1) * $tier + sort_min + $tier ))
     result=$(( $number <= $less_than && $number >= $greater_than ))

spark-test.sh

@@ -35,7 +35,7 @@ it_handles_decimals() {
   data="5.5,20"
   graph="$($spark $data)"
 
-  test $graph = '▂▇'
+  test $graph = '▃█'
 }
 
 it_charts_100_lt_300() {