Replace JSV parser

[willwhite]

We need a lightweight JSON schema parser that can run in the browser and allow us to provide custom error messages.

[miccolis]

@willwhite it looks like @Vertice may have tracked down a "better JSV" - https://github.com/Baggz/Amanda

He's going to be evaluating it the next day or so, but it looks promising.

[AdrianRossouw]

I wrote some tests in mocha to evaluate it, but I had to solve some pretty minor issues in amanda itself to get it to work.

I used the project schema and fixtures from tilemill to do the tests against, but I am not sure of it's completeness. The is no guarantee that when we write a constraint in our schemas they actually end up performing any or even the correct validation.

It's browser support also doesnt seem complete.

[AdrianRossouw]

One important wrinkle in the JSV implementation found in tilemill and other bones sites, is that unless you set an id on the model.schema, you will be unable to re-use a previously instantiated validation object and it will re-run the createSchema every time the validate function is called, which is could also be every time .set() is called.

Adding the 'id: "project"' line to the project schema should cause significant speed increases.

[AdrianRossouw]

While i am going through bones to find a general issue to the above problem, i looked into trimming down JSV a bit. You do not need the other environments to be sent to the client if you aren't validating against them, and the file sizes were as follows:

1. concat all requirements into one file - 115k
2. uglify.js - 44k
3. uglify + gzip - 11k

[yhahn]

Closing.

[springmeyer]

@yhahn's commit above (adding ID's to each schema) seems unrelated to this issue, and we are still using JSV, so re-opening.