"Time To Crack" is misleading

[crwatkins]

I noticed that 48 bits of entropy entered from Hex displays a Time To Crack of "less than a second" while 46 bits of dice throws displays "centuries". That doesn't make any sense (in a number of ways). A quick look at the code suggests that zxcvbn (which I'm a big fan of) is used to generate the Time to Crack. I believe the problem here is that zxcvbn is a password strength estimator, and not a generic entropy strength estimator. I believe a more accurate Time to Crack might be generated from the tool's entropy estimation.