Hi,
while working through Book 1 I opted for slightly different approach when applying fuzzy reflection in metal material.
Instead of modifying reflected ray direction I modify the hitpoint normal. On metal it does not make big difference (if any) but for me it was more intuitive to think about surface roughness by changing the normal of the surface at the hitpoint.

But what this also enabled me to do is to have fuzzy parameter for the dielectric, resulting in frosted glass outcome which I found neat