When the temperature starts to go above the human body temperature, most humans cannot survive in those environments. Thus, they would have little reason to describe such a temperature. Celsius wastes many double digit numbers between 40-100 that are rarely used. Instead, it forces you to use more negative numbers.

This winter, many days were in the 10s and 20s where I live. Using Celsius would have been marginally more inconvenient in those scenarios, which happen every winter. This is yet another benefit of Fahrenheit, it has a set of base 10 divisions that can be easily communicated, allowing for a convenient level of uncertainty when describing a temperature.

Generally -40 to 40 are the extremes of livable areas.

Sure, water is a really good system and it works well.

And for F that range is -40 to 104. See how you get 64 extra degrees of precision and nearly all of them are double digit numbers? No downside.

Furthermore F can use its base 10 system to describe useful ranges of temperature such as the 20s, 60s, etc. So you have 144 degrees instead of just 80, and you also have the option to utilize a more broad 16 degree scale that’s also built in.

You might say that Celsius technically also has an 8 degree scale(10s, 30s), but I would argue that the range of 10 degrees Celsius is too broad to be useful in the same way. In order to scale such that 0C is water freezing and 100C boiling, it was necessary for the units to become larger and thus the 10C shorthand is much less descriptive than the 10F shorthand, at least for most human purposes.