* Start refactoring kernel a bit and import some changes from kernel decoupling PR
* kernel: Put output always at the start in Syscall functions
* kernel: Rewrite GetThreadContext3 to use a structure and to be accurate
* kernel: make KernelTransfer use generic types and simplify
* Fix some warning and do not use getters on MemoryInfo
* Address gdkchan's comment
* GetThreadContext3: use correct pause flag
