The new API lends itself toward reading directly into the output buffer, saving
a memmove and allowing for easy reuse of output buffers.

This eliminates about half of the allocations by bytes for gcsfuse's sequential
read benchmark.

This accounts for ~80% of gcsfuse allocations.

The new interface gives much much cleaner documentation for package
fuseops, which is what we mostly look at. In particular, there isn't a
ton of stuttering in the method listings.

The change also simplifies the implementation of package fuseops, which
was becoming super unwieldy.