Minor bugfix.

Bug fixes:

• Use correct feature test macro for std::*_hardware_interference_size.

Bugfix

• Bug fix shared_from_this logic so it is actually callable.

Bugfix

• bug fix dangling reference returned by delay_gc_impl_::singleton().
• minor enhancement to generation shifting logic, should (in theory, under specific conditions) have slightly better performance).

Initial release

• It works! \o/
• Lots of documentation (because I like documenting things).
• Way too few test cases, but algorithm seems to work fine.