printing float values will have one more digit. (#13276) [backport]

* printing float values will have one more digit. Fixes #13196

tests/generics/tforwardgeneric.nim

@@ -1,11 +1,11 @@
 discard """
-  output: "1.1 11\n42\n0"
+  output: "1.125 11\n42\n0"
   ccodecheck: "!@'ClEnv'"
 """
 
 proc p[T](a, b: T): T
 
-echo p(0.9, 0.1), " ", p(9, 1)
+echo p(0.875, 0.125), " ", p(9, 1)
 
 proc p[T](a, b: T): T =
   let c = b
@@ -25,4 +25,3 @@ proc print[T](t: T) =
   echo t
 
 echo bar()
-

tests/generics/tgenerics_issues.nim

@@ -6,7 +6,7 @@ discard """
 perm: 22 det: 22
 TMatrix[3, 3, system.int]
 3
-@[0.9, 0.1]
+@[0.875, 0.125]
 U[3]
 U[(f: 3)]
 U[[3]]
@@ -20,9 +20,9 @@ concrete 88
 2
 3
 !!Hi!!
-G:0,1:0.1
-G:0,1:0.1
-H:1:0.1
+G:0,1:0.125
+G:0,1:0.125
+H:1:0.125
 '''
 joinable: false
 """
@@ -370,7 +370,7 @@ block t2304:
   type TV2[T:SomeNumber] = array[0..1, T]
   proc newV2T[T](x, y: T=0): TV2[T] = [x, y]
 
-  let x = newV2T[float](0.9, 0.1)
+  let x = newV2T[float](0.875, 0.125)
   echo(@x)
 
 
@@ -711,7 +711,7 @@ block t4863:
   proc q[j: static[int]](x:H[j]) = echo "H:", j, ":", x.v
 
   var
-    g0 = G[0,1](v: 0.1)
+    g0 = G[0,1](v: 0.125)
     h0:H[1] = g0
   p(g0)
   p(h0)