Support multiple cores in the runqueue.
The current allocation for each core is stored in a new array.
Thus, any call to `get_next_for_core` is minimal effort.
The allocation is updated after each change in the runqueue.

Export `CoreId`.

Clean code & improve readability.

Optimize case where `N_CORES` == 1.

Make hax (hopefully) happy.

In the multicore scenario, a yielding thread isn't necessarily the head
of the runqueue.
This commit adds a method `advance_from` to yield from any thread in the
queue. The thread is then moved to the tail of the runqueue.

Remove unused method `advance_head` (duplication of
`Runqueue::advance`).

Handle case N_CORES == 1 separately.

…to runqueue/multiple-cores

Add new `GlobalRunqueue` trait to differentiate and optimize for
different number of cores on the type level.

It uses the `min_specialization` rust feature to have a generic
implementation for _n_ number of cores and fine-grained specialization
of methods for concrete _n_.