Upgrade json-schema-to-typescript, update schema, switch to eslint

[eablack]

• Update json-schema-to-typescript to v14
• Update types to latest json.schema
• Replace tslint with latest eslint
• Introduce script for updating the types

[justinwilaby]

I have strong reservations about this. The ability to append arbitrary keys and make all properties optional to every interface within this schema is generally not useful.

This is a Postdeploy object according to this schema. It's also a PlatformAddOnPlanAction. It's also literally any type in this schema.

const actions: Postdeploy = { notThere: 'myValue' };

This usually indicates a mistake but TS allows it to go on it's merry way.

Also, this will mandate the use of the optional chaining operator when reading keys:

const postdeploy: Postdeploy = {exit_code: 0};
if (postdeploy?.exit_code === 0) {
  // Do stuff
}

PostDeploy.exit_code is also listed as readonly in the schema but it's optional/read/write in the interface

"exit_code": {
              "description": "The exit code of the postdeploy script",
              "example": 1,
              "readOnly": true,
              "type": [
                "integer"
              ]
            }

[justinwilaby]

I'm not sure this new schema alleviates the problems we have with the existing one.
1.readOnly flag not honored
2. strictProperties flag not honored. ref: https://github.com/java-json-tools/json-schema-validator/wiki/v5%3a-strictProperties
3. Strange unions and intersections with excessive inlining.

I believe we need a parser that understand JSON Hyper-Schema since this is what is specified in the Heroku API JSON document. e.g.

"release": {
      "description": "A release represents a combination of code, config vars and add-ons for an app on Heroku.",
      "$schema": "http://json-schema.org/draft-04/hyper-schema", <----------- Hyper Schema
// ...

[justinwilaby]

I'm not yet convinced this is the best path forward. See the main thread for a comment on my specific concerns.

[justinwilaby]

After a discussion with Eric, this PR will be used for 1.x while 2.x efforts focus on applying hyper-schema rules.