adds benchmark test to validate padding behaviour

[jinCN]

with a long type value,the implimentation of java Sequence will pad 7 bytes before the value and pad 7 bytes after the value,but why did you only pad 7 bytes after the value?

[fsaintjacques]

I'm trying to find the old source for the java disruptor. Also note that they're padding longs, not bytes.

[fsaintjacques]

See implemented test at http://mechanical-sympathy.blogspot.ca/2011/07/false-sharing.html

[fsaintjacques]

There's also an interesting point regarding this:

http://stackoverflow.com/questions/29199779/false-sharing-and-128-byte-alignment-padding