Use of uid2

I've seen in your examples you mostly use uid2 for unique id generation which is now (in version 0.0.3) using node's `crypto.pseudoRandomBytes()`. As stated in the documentation, these ids are [not necessarily unpredictable](http://nodejs.org/api/crypto.html#crypto_crypto_pseudorandombytes_size_callback), which means not good for token generation.
I think uid2 should be replaced with something using `crypto.randomBytes()` to prevent mistakes.

Start by locating the examples and token-generation paths that use uid2, then review how uid2 obtains randomness in the referenced version. Confirm that token generation uses unpredictable random bytes rather than pseudo-random bytes, and verify the affected examples or tests if present.

Written by the indexing model from the issue text.