Shenandoah: re-organize saving/restoring machine state in assembler code

Current XMM/GPR saving code is all over the place.

And it manifests in failures like:

$ CONF=linux-x86-server-fastdebug make images run-test TEST=compiler/c1/Test6855215.java TEST_VM_OPTS="-XX:+UseShenandoahGC"

#
# Internal Error (/home/shade/trunks/jdk/src/hotspot/cpu/x86/assembler_x86.cpp:3047), pid=1427307, tid=1427311
# Error: assert(VM_Version::supports_sse2()) failed

V [libjvm.so+0x53f9e8] Assembler::movsd(Address, XMMRegisterImpl*)+0x168
V [libjvm.so+0x14647bd] save_xmm_registers(MacroAssembler*)+0x9d
V [libjvm.so+0x1465d8f] ShenandoahBarrierSetAssembler::load_reference_barrier(MacroAssembler*, RegisterImpl*, Address, ShenandoahBarrierSet::AccessKind)+0x91f
V [libjvm.so+0x14686ee] ShenandoahBarrierSetAssembler::load_at(MacroAssembler*, unsigned long long, BasicType, RegisterImpl*, Address, RegisterImpl*, RegisterImpl*)+0x27e
V [libjvm.so+0x10e8e65] MacroAssembler::load_mirror(RegisterImpl*, RegisterImpl*, RegisterImpl*)+0x125
V [libjvm.so+0x165bd62] TemplateInterpreterGenerator::generate_fixed_frame(bool)+0x242
V [libjvm.so+0x1661bbd] TemplateInterpreterGenerator::generate_normal_entry(bool)+0x4ed
V [libjvm.so+0x1655d0c] TemplateInterpreterGenerator::generate_method_entry(AbstractInterpreter::MethodKind)+0x1cc
V [libjvm.so+0x16564b0] TemplateInterpreterGenerator::generate_all()+0x780
V [libjvm.so+0x1652e8b] TemplateInterpreter::initialize_code()+0xfb
V [libjvm.so+0xca8d7d] interpreter_init_code()+0x1d