arc: add ulps file for math tests

Using non-generic ulps file reduces failure counts for
math tests. Generated with nsim emulator running
test-{double,idouble,float,ifloat} -u and
gen-libm-test.pl on the build host.

test/math/libm-test-ulps-arc

@@ -0,0 +1,144 @@
+# Begin of automatic generation
+
+# cos
+Test "cos (M_PI_6l * 2.0) == 0.5":
+idouble: 1
+Test "cos (M_PI_6l * 4.0) == -0.5":
+double: 2
+float: 1
+idouble: 2
+ifloat: 1
+
+# erf
+Test "erf (0.75) == 0.711155633653515131598937834591410777":
+double: 1
+idouble: 1
+Test "erf (1.25) == 0.922900128256458230136523481197281140":
+double: 1
+idouble: 1
+
+# erfc
+Test "erfc (4.125) == 0.542340079956506600531223408575531062e-8":
+double: 1
+idouble: 1
+
+# exp
+Test "exp (1) == e":
+double: 1
+idouble: 1
+
+# expm1
+Test "expm1 (0.75) == 1.11700001661267466854536981983709561":
+double: 1
+idouble: 1
+
+# hypot
+Test "hypot (-0.7, -12.4) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (-0.7, 12.4) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (-12.4, -0.7) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (-12.4, 0.7) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (0.7, -12.4) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (0.7, 12.4) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (12.4, -0.7) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+Test "hypot (12.4, 0.7) == 12.419742348374220601176836866763271":
+float: 1
+ifloat: 1
+
+# lgamma
+Test "lgamma (0.7) == 0.260867246531666514385732417016759578":
+double: 1
+float: 1
+idouble: 1
+ifloat: 1
+Test "lgamma (1.2) == -0.853740900033158497197028392998854470e-1":
+double: 1
+float: 2
+idouble: 1
+ifloat: 2
+
+# log10
+Test "log10 (0.75) == -0.124938736608299953132449886193870744":
+double: 1
+idouble: 1
+Test "log10 (e) == log10(e)":
+float: 1
+ifloat: 1
+
+# tgamma
+Test "tgamma (-0.5) == -2 sqrt (pi)":
+double: 1
+idouble: 1
+Test "tgamma (0.7) == 1.29805533264755778568117117915281162":
+double: 1
+float: 1
+idouble: 1
+ifloat: 1
+
+# Maximal error of functions:
+Function: "cos":
+double: 2
+float: 1
+idouble: 2
+ifloat: 1
+
+Function: "erf":
+double: 1
+idouble: 1
+
+Function: "erfc":
+double: 1
+idouble: 1
+
+Function: "exp":
+double: 1
+idouble: 1
+
+Function: "expm1":
+double: 1
+idouble: 1
+
+Function: "hypot":
+float: 1
+ifloat: 1
+
+Function: "lgamma":
+double: 1
+float: 2
+idouble: 1
+ifloat: 2
+
+Function: "log":
+float: 1
+ifloat: 1
+
+Function: "log10":
+double: 1
+float: 1
+idouble: 1
+ifloat: 1
+
+Function: "tan":
+double: 1
+idouble: 1
+
+Function: "tgamma":
+double: 1
+float: 1
+idouble: 1
+ifloat: 1
+
+# end of automatic generation