I can't give you a quantitative answer to your question but I'll try to give you a feel for how CFD has affected the creation of engines. I'm going to start by discussing air breathing engines as a proxy for rocket engines.
I'm an aero/astro engineer and so was my father (who worked on several air breathing engine programs). In my father's era (J79 -> F110 time frame), engineers would "design" an engine, put it together, and then start a test program to determine its performance. Often the engines didn't work all that well.
Whether the engine worked well or not was an art on top of the science used to put it together. Once an engine was found to perform well, the engineers more or less kept reusing the same engine core but used different fans & AB sections with it to create engines for other uses (e.g. J79).
As I started working in the late 80s / early 90s there was still some of that going on (F101 -> F110 by reusing the core but changing the fans). However, big engine companies became less reliant upon reusing old cores for new engine families.
Don't get me wrong. Engine companies DO still make new engines out of proven cores but the new engines they design work far better on average than the ones that they developed in earlier decades.
Whether CFD actually reduces testing, I can't say. But it does permit engine companies to have a much greater chance of developing high performance engines.
And to bring it full circle, rocket engine reuse was common in earlier decades. The Russian's Soyuz RD-107 rocket engines are direct descendants of V2 rocket engines. Even the US still uses the descendants of the now 55+ year old RL10 engine for the same reason the engine companies reused air breathing engine designs.
But with CFD, companies like SpaceX are now able to design new engines from scratch. These have performed well so far.