It's good that you called Apple out on the throttling issue (after 1-2 *seconds*, seriously??), but this really highlights the lack of the Cinebench R15 multicore endurance test - the graph of CPU performance over time that you include in literally every single Windows laptop review and that visually emphasizes any throttling.
You really need to find a similar CPU endurance test for MacOS, especially since Geekbench is notorious for only testing peak performance during bursts.
Something just doesn't quite track. Compared to the 16" Pro, you measure lower average power commensurate with throttling during the CB 24 run, but then a higher score ... and they are in theory the same CPU. I'm not sure what is going on, but this is very strange - several oddities in these results which you noted as well, behavior seemingly needed to be fixed by Apple.
Apple's new high-end chip M5 Max is once again also available for the compact MacBook Pro 14. We test the fastest version with the 40-core GPU, 128 GB RAM and 8 TB of blazing fast PCIe 5.0 storage. But the M5 Max might just be too powerful for the 14-inch MacBook Pro...