Update powerpc64 ULPs.

The last upstream ULP update before the branch, in

commit c6922934363f44b88250567f52036d8e9972c255
Author: Andreas Schwab <schwab@linux-m68k.org>
Date:   Sat Mar 3 14:20:24 2012 +0100

    Update powerpc libm ULPs

did not specify 1-ULP tolerance for all tests because some of them
returned the exact result.  Current upstream uses a different way for
specifying ULPs, after this commit:

commit e6b6a85705be16373fb3f78ef998edc62150499c
Author: Joseph Myers <joseph@codesourcery.com>
Date:   Wed Mar 5 15:02:38 2014 +0000

    Don't include individual test ulps in libm-test-ulps.

This implicitly increased the test tolerance for most (all?) math
functions to 1 ULP because at least one subtest was off by 1 ULP.

Index: b/sysdeps/powerpc/fpu/libm-test-ulps
===================================================================
--- a/sysdeps/powerpc/fpu/libm-test-ulps
+++ b/sysdeps/powerpc/fpu/libm-test-ulps
@@ -2396,6 +2396,9 @@ ldouble: 2
 Test "sin_downward (8) == 0.9893582466233817778081235982452886721164":
 ildouble: 1
 ldouble: 1
+Test "sin_downward (9) == 0.4121184852417565697562725663524351793439":
+float: 1
+ifloat: 1
 
 # sin_tonearest
 Test "sin_tonearest (1) == 0.8414709848078965066525023216302989996226":
@@ -2438,10 +2441,15 @@ float: 1
 ifloat: 1
 ildouble: 2
 ldouble: 2
+Test "sin_upward (10) == -0.5440211108893698134047476618513772816836":
+float: 1
+ifloat: 1
 Test "sin_upward (2) == 0.9092974268256816953960198659117448427023":
 float: 2
 ifloat: 2
 Test "sin_upward (3) == 0.1411200080598672221007448028081102798469":
+float: 1
+ifloat: 1
 ildouble: 1
 ldouble: 1
 Test "sin_upward (4) == -0.7568024953079282513726390945118290941359":
@@ -2449,9 +2457,18 @@ float: 1
 ifloat: 1
 ildouble: 1
 ldouble: 1
+Test "sin_upward (5) == -0.9589242746631384688931544061559939733525":
+float: 1
+ifloat: 1
 Test "sin_upward (6) == -0.2794154981989258728115554466118947596280":
 ildouble: 1
 ldouble: 1
+Test "sin_upward (7) == 0.6569865987187890903969990915936351779369":
+float: 1
+ifloat: 1
+Test "sin_upward (8) == 0.9893582466233817778081235982452886721164":
+float: 1
+ifloat: 1
 Test "sin_upward (9) == 0.4121184852417565697562725663524351793439":
 float: 1
 ifloat: 1