Fix compiling for linux arm64

[JeodC]

This pull request contains several changes.

1. In HMAS.cpp a typedef mismatch causes linux arm64 builds to fail. It has been altered in HMAS.h to pass.
2. In CMakeLists.txt there was no definition to enable BUILD_MK64 with the externally compiled torch, which would cause torch to throw an error on trying to make mk64.o2r. This definition has been added.
3. In macros.h ALIGNED16 and ALIGNED8 were empty. I added the alignment to the macros so they do something, and I added ALIGNED4096 so the 256MB allocation is set at the start of the virtual memory page.
4. Added the ALIGNED16 macro to the memory pool in memory.c
5. In CMakeLists.txt added USE_OPENGLES define
6. Fixed a rendering bug in LUS: fix(fast): resolve 64-bit pointer truncation in SegAddr for ARM64 Kenix3/libultraship#1004
7. Changed to memset

#490 will probably want to rebase on top of this