- Added a LICENSE file
 - Added a code generater to generate ES and bit size
	specific code.
The speedup of the generated code really depends. For some operations
its a lot faster. For sqrt it may seem slower, but this is because
allignment specific things that impact the tight loop. The non-generated
code just happens to be more lucky then the generated code. This should
not affect real world performance as allignment is impacted by a lot of
variables