Fix BaseDomain method ordering for row vectors

Since IsMatrix has a high rank and IsGF2VectorRep / Is8BitVectorRep objects
are in the filter IsMatrix, the wrong BaseDomain method was selected for such
objects. This reduces performance, and also caused an error with vectors of
length 0.

lib/matrix.gi

@@ -1258,6 +1258,7 @@ end );
 InstallOtherMethod( BaseDomain,
     "generic method for a row vector",
     [ IsRowVector ],
+    -SUM_FLAGS,
     DefaultRing );
 
 InstallOtherMethod( OneOfBaseDomain,

tst/testinstall/MatrixObj/ZeroVector.tst

@@ -5,9 +5,9 @@ gap> ReadGapRoot("tst/testinstall/MatrixObj/testmatobj.g");
 # IsGF2VectorRep
 #
 gap> TestZeroVector(IsGF2VectorRep, GF(2), 3);
-Error, Assertion failure
+<a GF2 vector of length 3>
 gap> TestZeroVector(IsGF2VectorRep, GF(2), 0);
-Error, Assertion failure
+<a GF2 vector of length 0>
 
 # test error handling
 gap> TestZeroVector(IsGF2VectorRep, GF(2), -1);
@@ -19,15 +19,15 @@ Error, IsGF2VectorRep only supported over GF(2)
 # Is8BitVectorRep
 #
 gap> TestZeroVector(Is8BitVectorRep, GF(3), 3);
-Error, Assertion failure
+[ 0*Z(3), 0*Z(3), 0*Z(3) ]
 gap> TestZeroVector(Is8BitVectorRep, GF(3), 0);
-Error, Z: <q> must be a positive prime power (not the integer 1)
+< mutable compressed vector length 0 over GF(3) >
 
 #
 gap> TestZeroVector(Is8BitVectorRep, GF(251), 3);
-Error, Assertion failure
+[ 0*Z(251), 0*Z(251), 0*Z(251) ]
 gap> TestZeroVector(Is8BitVectorRep, GF(251), 0);
-Error, Z: <q> must be a positive prime power (not the integer 1)
+< mutable compressed vector length 0 over GF(251) >
 
 # test error handling
 gap> TestZeroVector(Is8BitVectorRep, GF(3), -1);