This resolves #477

I reused the same function with an additional argument with default value for backward compatibility (compared to the current master).
I prefer that to a separate function, but let me know if you think otherwise.

Only the specific axi_stream stuff is updated. One could think of doing the same to the generic stream_slave, but I can't handle the consequences (would need to update the uart...).

Running all checks in verification_components seems to fail with questa due to some delta-cycle differences with GHDL... :-/ GHDL looks fine here.