This adds support for the Databricks CLI.

This CLI needs to know a "host" URL to the instance of Databricks you're working with, as well as your credentials that are specific to that instance. Personal access tokens are they primary way to authenticate, while username+password is possible but discouraged.

I've tested this on my own config, and both import and usage seems to work just fine 💯 Please let me know if you'd like to see more test cases or anything else before accepting.

A couple notes:

• I've opted for the "Person Access Token" credentials type, despite username+password being an option. The code here handles both cases, but the label of "Personal Access Token" isn't quite right in both cases. I still think it should be as I've proposed here to encourage the recommended flow, but an alternative would be to use a more generic "Configuration" label. A second opinion here would be nice 🙂
• When doing op plugin list, this plugin shows up with REQUIRED FIELDS: host. This is because while you do need to provide either a token or a username and password, neither of them can be marked as required on their own. Probably not a big deal, but resolving it can't really be done without making some changes to schema.CredentialField and related code.
• The config files at the moment typically also includes a jobs-api-version = 2.x line. This is ignored by the plugin with the following reasoning:
  1. Unlike the other config options, the CLI does not accept setting the API version with an environment variable. In order to support this the plugin would have to make use of the file-based provisioning variant, as well as add a highly specific field definition to the SDK. While possible, I don't believe it is justified given the next two points.
  2. The version can be set with a command line flag, and is only relevant for the databricks jobs subcommand.
  3. Perhaps most importantly, this version flag is temporary, and future versions of the CLI will default to the newer API version.