Add partial support for future values of LOGICAL_PROCESSOR_RELATIONSHIP

[dbwiddis]

See #1324 for more details.

This PR implements new LOGICAL_PROCESSOR_RELATIONSHIP enum values which are present in current Windows Insider builds and currently throw an IllegalStateException when instantiating a SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX structure from a pointer.

A complete implementation will require updating the NUMA_NODE_RELATIONSHIP structure and javadocs, and will be submitted later once the new API documentation is available.

[matthiasblaesing]

I admit I have only glanced over, but just a thought - would it make sense to return a generic value holding the raw memory value when an unsupported relationship is encountered? That way people would at least get a value and not blow with an exception - especially as it is now proven, that new relationships are possible. Possible name "UNKNOWN_RELATIONSHIP".

[dbwiddis]

Possible name "UNKNOWN_RELATIONSHIP".

Heh. I actually had previously implemented a similar workaround with exactly that name and am happy to go that route. My reasoning to not submit it that way was that after the new code was in we'd never use that structure again.

We'll still have the two leading int values for relationship and size; what would you suggest to retrieve the remaining data? I'm thinking no field, but a convenience getter that returns a Pointer sharing the "size" number of bytes of memory.

[dbwiddis]

Actually such a getter isn't needed, the structure itself already has a getPointer() that returns the same thing.

[matthiasblaesing]

Looks sane to me.