Add greater time spans in display, Add 1e12 guess metric

Add larger time spans

The textual feedback (result.crack_times_display) on times stops at centuries. Considering Moore's law and assuming guessing speed progresses proportionally, a century long password today will take 3 years in a decade and 35 days in two decades.

In that vein, I added millennium (1000 years) its associated pluralization. If a password exceeds 1000 millennia in guessing time (a million years), then the returned value is ">1000 millennia". This draws a better distinction between currently-strong passwords and forever-strong passwords.

Add 1e12 guesses

Leaving guess counts at 1e10 seems a little optimistic, considering we know of organizations with the ability to guess a trillion or more times a second (the README says as much) and home-made rigs have long since broken 330GH using Hashcat. As such, I added 1e12 guesses to the result for more enthusiastic password security.

Why should this matter?

Note that if the two points made above are combined (exponential growth in guessing power and underestimation of guessing power), the following conclusions can be made:

Take a password that is rated to take a millennium to crack against a 1e10 / sec guesser.
If pitched against a 1e12 / sec guesser, that password drops to 10 years.
After a decade of technological advancement, the same password drops to about 100 days of guessing time against a commensurate attacker.
After two decades, that drops to about 8 hours against a commensurate attacker.

Concerns about scope

I understand if this seems unnecessary to the scope of the project.
This change is very small. If it's deemed unnecessary to merge it, I'm fine with keeping it that way.

Testing

This code makes no changes to the way zxcvbn works, it only adds to the textual output. As such, npm test passes all 1027 tests.

dropbox / zxcvbn