Add option to generate mnemonic from camera feed

[ghost]

It looks like SeedSigner does this, and it seems like a promising way to generate a mnemonic that is random if done right.

Will need to brush up on my stats knowledge, but here's my initial rough plan:

I'm thinking you would calculate entropy from the images using Shannon's formula and display to the user a progress bar / percent until enough entropy is collected. Once you have good/enough entropy, the words (numbers, really) are generated and then a Chi-squared test is performed on the generated words as a sanity check.

May want to also use a similarity function to compare the images and reject "duplicates" so that the user is forced to move the camera around?