Update test configs

mega-test-quick.cfg

-linux-gcc-4.7      1900 ./super-test.sh tmpdir capnp-gcc-4.7 quick
-linux-gcc-4.8      1903 ./super-test.sh tmpdir capnp-gcc-4.8 quick gcc-4.8
-linux-clang        1935 ./super-test.sh tmpdir capnp-clang quick clang
+linux-gcc-4.7       930 ./super-test.sh tmpdir capnp-gcc-4.7 quick gcc-4.7
+linux-gcc-4.8       929 ./super-test.sh tmpdir capnp-gcc-4.8 quick gcc-4.8
+linux-clang         959 ./super-test.sh tmpdir capnp-clang quick clang
 mac                 884 ./super-test.sh remote beat caffeinate quick
-cygwin              809 ./super-test.sh remote Kenton@flashman quick
+cygwin              923 ./super-test.sh remote Kenton@flashman quick

mega-test.cfg

-linux-gcc-4.7     14779 ./super-test.sh tmpdir capnp-gcc-4.7
-linux-gcc-4.8     13122 ./super-test.sh tmpdir capnp-gcc-4.8 gcc-4.8
-linux-clang       14907 ./super-test.sh tmpdir capnp-clang clang
-mac                5741 ./super-test.sh remote beat caffeinate
-cygwin             6764 ./super-test.sh remote Kenton@flashman
+linux-gcc-4.7      5952 ./super-test.sh tmpdir capnp-gcc-4.7 gcc-4.7
+linux-gcc-4.8      5246 ./super-test.sh tmpdir capnp-gcc-4.8 gcc-4.8
+linux-clang        6039 ./super-test.sh tmpdir capnp-clang clang
+mac                5758 ./super-test.sh remote beat caffeinate
+cygwin             6770 ./super-test.sh remote Kenton@flashman

super-test.sh

@@ -72,6 +72,9 @@ while [ $# -gt 0 ]; do
     gcc-4.8 )
       export CXX=g++-4.8
       ;;
+    gcc-4.7 )
+      export CXX=g++-4.7
+      ;;
     kenton )
       cat << "__EOF__"
 =========================================================================
@@ -222,14 +225,12 @@ if [ $IS_CLANG = yes ]; then
   # There's an unused private field in gtest.
   export CXXFLAGS="$CXXFLAGS -Wno-unused-private-field"
 else
-  if (${CXX:-g++} --version | grep -q ' 4[.]8'); then
-    # GCC 4.8 emits a weird uninitialized warning in kj/parse/char-test, deep in one of the parser
-    # combinators.  It could be a real bug but there is just not enough information to figure out
-    # where the problem is coming from, because GCC does not provide any description of the inlining
-    # that has occurred.  Since I have not observed any actual problem (tests pass, etc.), I'm
-    # muting it for now.
-    CXXFLAGS="$CXXFLAGS -Wno-maybe-uninitialized"
-  fi
+  # GCC 4.8 emits a weird uninitialized warning in kj/parse/char-test, deep in one of the parser
+  # combinators.  It could be a real bug but there is just not enough information to figure out
+  # where the problem is coming from, because GCC does not provide any description of the inlining
+  # that has occurred.  Since I have not observed any actual problem (tests pass, etc.), I'm
+  # muting it for now.
+  CXXFLAGS="$CXXFLAGS -Wno-maybe-uninitialized"
 fi
 
 cd c++