Clean up validation outputs

Currently, the outputs of validation is raw mocha results, it's a bit verbose.

Like,

```
21) /repos/{ownerId}/{repoId}/milestones POST response 201:
AssertionError: Got unexpected response body:
{
"url": "https://api.github.com/repos/octocat/Hello-World/milestones/1",
"number": 1,
"state": "open",
"title": "v1.0",
"description": "",
"creator": {
"login": "octocat",
"id": 1,
"avatar_url": "https://github.com/images/error/octocat_happy.gif",
"gravatar_id": "somehexcode",
"url": "https://api.github.com/users/octocat"
},
"open_issues": 4,
"closed_issues": 8,
"created_at": "2011-04-10T20:09:31Z",
"due_on": null
}
Error: expected value to match json-schema '{"$schema":"http://json-schema...' -> '{"url":"https://api.github.com...'
invalid type: null (expected string)
schema: {"$schema":"http://json-schema.org/draft-03/schema","type":"...
rule: /properties/due_on/type -> "string"
field: /due_on -> null: null
at Function.assert.jsonSchema (/usr/local/lib/node_modules/ramlev/node_modules/chai-json-schema/index.js:176:40)
at Test.assertExample (/usr/local/lib/node_modules/ramlev/lib/test.js:77:21)
at Test.assertExample (/usr/local/lib/node_modules/ramlev/lib/test.js:3:61)
at Test.run (/usr/local/lib/node_modules/ramlev/lib/test.js:56:21)
at Object.suite.addTest.Mocha.Test._.bind.test (/usr/local/lib/node_modules/ramlev/lib/test-runner.js:37:19)
at Test.Runnable.run (/usr/local/lib/node_modules/ramlev/node_modules/mocha/lib/runnable.js:217:15)
at Runner.runTest (/usr/local/lib/node_modules/ramlev/node_modules/mocha/lib/runner.js:373:10)
at /usr/local/lib/node_modules/ramlev/node_modules/mocha/lib/runner.js:451:12
at next (/usr/local/lib/node_modules/ramlev/node_modules/mocha/lib/runner.js:298:14)
at /usr/local/lib/node_modules/ramlev/node_modules/mocha/lib/runner.js:308:7
```

Places can improve:
- [ ] Remove error stack
- [ ] Locate the line # of invalid example in RAML
- [ ] Better error output for json

Start by running the RAML validation that produces the raw Mocha output shown in the issue, then trace where that output is assembled. Compare the current output with each checklist item: remove the stack, identify the invalid example's RAML line, and improve JSON-schema error formatting. Done means validation failures are shorter and provide the requested context.

Written by the indexing model from the issue text.