fix: reading dirty memory from Util::HasOnlyZeros in G2Element deserialization

[knst]

It fixes this bug while running with valgrind:

==4670== Conditional jump or move depends on uninitialised value(s)
==4670==    at 0x1041396: bls::G2Element::FromBytesUnchecked(bls::Bytes, bool) (in DASH/src/test/test_dash)
==4670==    by 0x10417C6: bls::G2Element::FromBytes(bls::Bytes, bool) (in DASH/src/test/test_dash)
==4670==    by 0x43A735: CBLSWrapper<bls::G2Element, 96ul, CBLSSignature>::SetByteVector(std::vector<unsigned char, std::allocator<unsigned char> > const&, bool) (bls.h:113)
==4670==    by 0x43A971: void CBLSWrapper<bls::G2Element, 96ul, CBLSSignature>::Unserialize<CDataStream>(CDataStream&, bool, bool) (bls.h:190)
==4670==    by 0x43111E: Unserialize<CDataStream> (bls.h:210)
==4670==    by 0x43111E: Unserialize<CDataStream, CBLSSignature&> (serialize.h:906)
==4670==    by 0x43111E: operator>><CBLSSignature&> (streams.h:465)
==4670==    by 0x43111E: bls_tests::FuncSerialize(bool) (bls_tests.cpp:50)
==4670==    by 0x431C5F: test_method (bls_tests.cpp:418)
==4670==    by 0x431C5F: bls_tests::bls_serialize_tests_invoker() (bls_tests.cpp:416)
==4670==    by 0x115D4E1: boost::detail::function::function_obj_invoker0<boost::detail::forward, int>::invoke(boost::detail::function::function_buffer&) (in DASH/src/test/test_dash)
==4670==    by 0x115BB8C: boost::execution_monitor::catch_signals(boost::function<int ()> const&) (in DASH/src/test/test_dash)
==4670==    by 0x115BC14: boost::execution_monitor::execute(boost::function<int ()> const&) (in DASH/src/test/test_dash)
==4670==    by 0x115BCF4: boost::execution_monitor::vexecute(boost::function<void ()> const&) (in DASH/src/test/test_dash)
==4670==    by 0x1122824: boost::unit_test::unit_test_monitor_t::execute_and_translate(boost::function<void ()> const&, unsigned long) (in DASH/src/test/test_dash)
==4670==    by 0x10FC17F: boost::unit_test::framework::state::execute_test_tree(unsigned long, unsigned long, boost::unit_test::framework::state::random_generator_helper const*) [clone .isra.0] (in DASH/src/test/test_dash)
==4670== 

[ogabrielides]

LGTM, but why some CI fail?

[knst]

but why some CI fail?

@ogabrielides , seems as CI is broken for awhile, previous PRs fails same:

• feat: serialize on the stack #75
• fix: dash is no more pretending being a 'libdashbls' #76

I haven't check why yet, but it seems as all of them is Go's related.

[knst]

@ogabrielides here's PR from Konstantin to fix CI failure. #82
Please, help to merge this one.

[UdjinM6]

utACK

[UdjinM6]

utACK (preferably after #82 + rebase)