Added implementation of multi-tail circular buffer with O1 head space
allocation.
This intended for situation, where more asynchronous sources
need to write data into queue. Because of uLUt is intended
to not enforce any policy to its users, the locking is not
part of the code and operations are divided to the shorter sequences.
Some of these has to be called with lock, spinlock or even IRQ
protection. Synchronization selection is fully in client hands.