big: Rat always outputs the requested precision from FloatString

Fixes #1922.

R=golang-dev, gri
CC=golang-dev
https://golang.org/cl/4551098

src/pkg/big/rat.go

@@ -314,7 +314,11 @@ func (z *Rat) RatString() string {
 // digits of precision after the decimal point and the last digit rounded.
 func (z *Rat) FloatString(prec int) string {
 	if z.IsInt() {
-		return z.a.String()
+		s := z.a.String()
+		if prec > 0 {
+			s += "." + strings.Repeat("0", prec)
+		}
+		return s
 	}
 
 	q, r := nat{}.div(nat{}, z.a.abs, z.b)

src/pkg/big/rat_test.go

@@ -86,12 +86,13 @@ var floatStringTests = []struct {
 	out  string
 }{
 	{"0", 0, "0"},
-	{"0", 4, "0"},
+	{"0", 4, "0.0000"},
 	{"1", 0, "1"},
-	{"1", 2, "1"},
+	{"1", 2, "1.00"},
 	{"-1", 0, "-1"},
 	{".25", 2, "0.25"},
 	{".25", 1, "0.3"},
+	{".25", 3, "0.250"},
 	{"-1/3", 3, "-0.333"},
 	{"-2/3", 4, "-0.6667"},
 	{"0.96", 1, "1.0"},