Store shuffled alphabets so that they don't have to be calculated again.

[jwpage]

This changes the Hashids::bench function so that it will store a shuffled alphabet once it's been created. This increases the performance of the function, especially when calculating large numbers of hashes.

Here are a couple of blackfire.io profiles for encoding 50,000 hashes.

• Before: 8.99s
• After: 4.96s

[vinkla]

Thanks for the pull request @jwpage! This pull request looks good to me.

@ivanakimov what do you say?

[jd327]

@vinkla @jwpage this is super awesome.

@jwpage thanks for providing benchmarks as well! I was trying to think of any scenarios where this might break anything, but I guess besides memory consumption if the tests are good, this should be a nice optimization.

@vinkla if looks good to you, let's get it out!

[vinkla]

Merged! This will be part of version 2.1 which will be released in the coming weeks. For now, you could set minimum-stability to dev and start using version 2.1 in development today.