Like I said -- we're limited by the speed of light in distributing blocks far more than we are in machine processing time of blocks. Block processing is in the order high micro- to millisecond, where block propagation around the world is high milli- to full second. There's plenty of time to forgive processing language inefficiency. I say this with confidence as a) a computer engineer, and b) a witness here on steem. Latency is the biggest issue for consensus!
If you want to reeaaaally did deep in to what takes longer on python compared to C, you'll find a second non-issue. The stuff that takes the longest amount of time in block processing is transaction verification, which is cryptography. Guess what? Python can call functions in libraries that are actually implemented in C. So, for the biggest bottleneck, you'll get the almost the exact same performance.