Changed tests in testinstall that did rely on immediate methods

Ditto for manual examples, including
some stability (do not test for values of `Random` etc.)

Includes slight modification for DerivedSubgroup of fp group
if subgroup of finite index is cyclic. (This might look overly specialized,
but is apparently of interest for some test examples, in any case it comes
at little cost to other calculations.)

also Further github remarks.

Fixed algmat.tst:

It seems this test file is run in different configurations, in which a
basis for a 0-dim space in one example is considered SemiEchelon (and thus
prints differently), in the other case not. Thus remove the display.
(Another example why the concept of testing for verbatim output is flawed.)

Another manual example changed due to further deductions

New trace.out from master

Changed tests in `testinstall` that did rely on immediate methods

Ditto for manual examples, including
some stability (do not test for values of `Random` etc.)

Includes slight modification for DerivedSubgroup of fp group
if subgroup of finite index is cyclic. (This might look overly specialized,
but is apparently of interest for some test examples, in any case it comes
at little cost to other calculations.)

also Further github remarks.

Fixed algmat.tst:

It seems this test file is run in different configurations, in which a
basis for a 0-dim space in one example is considered SemiEchelon (and thus
prints differently), in the other case not. Thus remove the display.
(Another example why the concept of testing for verbatim output is flawed.)

Another manual example changed due to further deductions

New trace.out from master

tst/test-error/trace.g.out

@@ -157,3 +157,4 @@ gap> o(IsInt,2,3,4,5,6);
 gap> o(IsInt,2,3,4,5,6,7); # not (yet?) supported
 Error, sorry: cannot yet have X argument constructors
 not in any function at *stdin*:94
+gap> QUIT;