math/big: remove NaN support - just not worth it

NaNs make the API more complicated for no real good reasons.
There are few operations that produce NaNs with IEEE arithmetic,
there's no need to copy the behavior. It's easy to test for these
scenarios and avoid them (on the other hand, it's not easy to test
for overflow or underflow, so we want to keep +/-Inf).

Also:
- renamed IsNeg -> Signbit (clearer, especially for x == -0)
- removed IsZero           (Sign() == 0 is sufficient and efficient)
- removed IsFinite         (now same as !IsInf)

Change-Id: I3f3b4445c325d9bbb1bf46ce2e298a6aeb498e07
Reviewed-on: https://go-review.googlesource.com/8280
Reviewed-by: Alan Donovan <[email protected]>

src/math/big/accuracy_string.go

@@ -4,13 +4,14 @@ package big
 
 import "fmt"
 
-const _Accuracy_name = "ExactBelowAboveUndef"
+const _Accuracy_name = "BelowExactAbove"
 
-var _Accuracy_index = [...]uint8{0, 5, 10, 15, 20}
+var _Accuracy_index = [...]uint8{0, 5, 10, 15}
 
 func (i Accuracy) String() string {
+	i -= -1
 	if i < 0 || i+1 >= Accuracy(len(_Accuracy_index)) {
-		return fmt.Sprintf("Accuracy(%d)", i)
+		return fmt.Sprintf("Accuracy(%d)", i+-1)
 	}
 	return _Accuracy_name[_Accuracy_index[i]:_Accuracy_index[i+1]]
 }