Unfortunately, no. The root cause(s) have been speculated on, and the capacitor between EN / GND only helps a bit (reboot is more reliable, but slower).
To find the actual root cause(s) could be a significant effort, involving oscilliscopes with triggers, and many hours of conjecture + testing loops. The cost is too high, except perhaps for EE college students. Even then, it’s a bit of a “whack-a-mole” … you might think you’ve found all the causes, but do you really want to save $20/board, only to spend an entire weekend trying to figure out why it’s randomly failing?
People have looked at the easy stuff … the low-hanging fruit. These were inexpensive for a reason … they were single-purpose ethernet-to-serial converters, which are inherently low-CPU, low-power. What if they were inexpensive b/c they got 3rd-quality chips that auto-reset due to thermal issues when running for more than 100ms? It’s just not possible to know.
On the other hand, if you have the means, the knowledge, and the time to determine an inexpensive fix for these devices via deep-dive analysis, that’s AWESOME! Please share your findings!