Feature: Models v2

[timotheeguerin]

fix #162 (Unkown keys in config files will raise error)
Define a new Model base class and fields to be able to define models as below
This will handle most of the Validation, merging, parsing from dict automatically

class UserState(Enum):
    Creating = "creating"
    Ready = "ready"
    Deleting = "deleting"

class UserInfo(Model):
    name = fields.String()
    age = fields.Integer()

class User(Model):
    info = fields.Model(UserInfo)
    enabled = fields.Boolean(default=True)
    state = fields.Enum(UserState, default=UserState.Ready)

# This is a valid way to create(from dict)
user = User(
        info=dict(
            name="Highlander",
            age=800,
        ),
        enabled=False,
        state="deleting",
    )

Todo:

• Handle list merging(Allow append, replace options)
• Handle nested object merging(Allow merge, replace options)
• Handle list parsing and validation correctly
• Apply on existing models using configuration base
• update writing model docs

[darylmsft]

is true the production value?

[timotheeguerin]

This setting is to have option to run/debug the test one by one in vscode instead of using the cli, this is just for development

[jafreck]

why are some models left in this file? shouldn't this be in models/file_share.py now?

[jafreck]

Not sure if this applies to this PR or if it needs another, but shouldn't this file exist in the aztk.spark.models namespace since this has all of the spark plugins specified?

[jafreck]

Shouldn't we have implementations for all defined types of ListMergeStrategy?

[timotheeguerin]

The other one is replace which is the default

[jafreck]

for cases like this, should we be handling for a KeyError and returning something nicer if vm_count doesn't exist, for example?

[jafreck]

Also, can we change k to key?

[jafreck]

what happens if size is set to 0 and size_low_pri is set to None?

[jafreck]

I feel like this is a little confusing. Maybe we should point to docs on how to format secrets instead?