consider using a ring-buffer instead of an array

the implementation of the FifoLogger uses an array with push, slice and splice methods.
it seems to me that implementing it with a fixed-sized ring buffer could be more efficient.

Locate the FifoLogger implementation and inspect how its array, push, slice, and splice operations define current FIFO behavior. Determine the fixed capacity and benchmark or test the existing behavior before replacing the storage approach; done means equivalent logging behavior with the proposed ring buffer and evidence of an efficiency improvement.

Written by the indexing model from the issue text.