Load Store optimization and initial Clang support

• Added initial support for Clang. Now Clang can produce better IR code when it knows what the arch is capable of
• Added LoadStore Optimization pass. It enables use of wide loads and stores when it is available.

Extended 64-bit and FPU/IALU2 support

With this release, more complex code can run, such as one mixing FPU and IALU2 instructions, using inline asm and 64-bit values and branches.

Changelog:

• Fixed FP handling
• Fixed branch lowering
• Fixed some of 64-bit issues
• Fixed asm parser
• Reworked FPU/IALU2 config flag setting pass

Initial release

First more or less working version.

Known issues:

• 64-bit regs broken when dealing with external libs because of wrong order
• Frame pointer elimination broken
• Asm parser broken
• Might be some issues with branching