Endianness

[BeckySharp]

PLAN:

• save endianness of index in buildinfo or somewhere, where we can check at the beginning
• when launching odinson, check native endianness
• based on compatibility, when initializing the unsafe serializer, decide which version of the get/put methods to use, set (have 2 versions of serializer methods -- one that reverses the bytes and one that doesn't)

ALSO:

• have "validation" program that checks things like endianness but maybe more
• if endianness is not compatible, suggest user reindex if odinson docs/corpus available to have fastest runtime possible

OTHER:

• can we update a field?

FYI @kwalcock thanks for the thoughts in processors land, do you have thoughts on this?

[kwalcock]

The default should probably be whatever the Intel chips have so that there is no need to record it in the file, IMO. If a different endianness is found, the bytes can be swapped and swapped back. Those machines will just be slightly (I assume) slower. If the memory layout can vary, there are more than two possibilities, but that's going too far.

If this is the case "The Intel x86 and AMD64 / x86-64 series of processors use the little-endian format." and this is the case "Internally, Java uses Big Endian." then we should be able to look at the output and notice the contradiction---and make sure I'm not crazy.