Refactor Skeleton interface

[phorward]

Ok, this will become a huger and recently updated issue relating to several things in VIURs current Skeleton interface. The reason for this issue is, that the current interface is uneven, messy and misleading in several ways.

CRUD-functions

It's called CRUD: Create, Read, Update, Delete.

Current state

For Skeletons, this is:

• Create: toDB() with skel["key"] set to None
• Read: fromDB()
• Update: toDB() with a skel["key"] set
• Delete: delete()

Okay - why is it fromDB(), toDB() but not... deleteDB()? This is uneven and messy.

Proposed state

• Create: write() with skel["key"] set to None
• Read: read()
• Update: write() with a skel["key"] set
• Delete: delete()

Hooks

In many situations, it is necessary to do some pre- or post-processing of data before it is written, read or deleted.

Current state

• toDB():
  • preProcessSerializedData() - Do something on the already serialized db.Entity without using the bones anymore (mostly useless!)
  • preProcessBlobLocks() - Use is only relevant to blobs?
  • postSavedHandler() oh nice, I can do something after the skeleton has been saved. Why isn't it called postToDBHandler()?
  • Ok how to use the bones preliminary before writing data? Right, you override toDB(), do your preparation and afterwards make a super call. This is the current way to go.
• fromDB(): No chance to do something here, except overriding fromDB(). Wow.
• delete():
  • postDeletedHandler() is a hook on a SkeletonInstance which has been previously deleted.

Proposed state

It should be equivalent to module's hooks functionalities and naming:

• write():
  • on_write() - do something before saving with the skeleton
  • serialize() - do something before saving with the db.Entity
    default/super should be a Skeleton.serialize(), which doesn't exist as well!
• read():
  • unserialize() - do something after reading with the db.Entity
Skeleton.unserialize() doesn't exist anymore, as bones are unserialized on-demand. Anyway, the hook should exist for override.
  • on_read() - do something after reading with the skeleton
• delete():
  • on_delete() - do something after deleting the dataset. This is mostly equal to postDeletedHandler(), except the signature.

Shortcut functions

The following shortcut functions should be made available for fast skeleton CRD (create-read-delete).

Skeleton.get()

It's should be possible to just load a skeleton by a given key immediately.

Something like

article = skeleton.Skeleton.get(" ahBteS1mdW5ueS1wcm9qZWN0ciMLEgdhcnRpY2xlIhZteS1zZWNyZXQtYXJ0aWNsZS1oYWhhDA")

This could be achieved by a staticmethod

class Skeleton:
    @staticmethod
    def get(key: db.Key | str | int, kind: Option[str] = None) -> SkeletonInstance
        # ...

• key: can be either a db.Key or an encoded key, or a name or an int. Latter one will only be used when kind is set. On kind mismatch, an exception is raised.
• kind: Optionally defines the kind of the db.Entity, to determine the Skeleton. In case the kind is not provided, it is extracted from either the db.Key or the encoded-str, otherwise the kind is mandatory.

The function shall return None in case the given key wasn't found.

Skeleton.new()

tbd

Skeleton.del()

tbd