Adding inline M2M fields

[joewesch]

First off, I'm sorry for the extra lengthy PR. This has been a personal hackathon project that I've been working on for the last month or so and I think it's going to be something many people will want.

This PR introduces a framework for implementing inline M2M fields on parent modules. This replaces the need to separately create (for example) clusters, devices and device to cluster associations. Now, you can create the cluster and then associate it to the device as part of the device module:

  networktocode.nautobot.device:
    ...
    clusters:
      state: merge
      objects:
        - cluster: "Test Cluster 2"

You can also now create Select and Multi-Select custom fields with the choices all with one module:

  networktocode.nautobot.custom_field:
    ...
    type: select
    custom_field_choices:
      state: replace
      objects:
        - value: "Choice A"
          weight: 100

Each of these M2M associations support their own independent state: merge (default), replace or delete. The rest of the module remains 100% the same and backwards compatible.

As part of this PR, I have deprecated all existing modules that solely made these M2M associations. Once this is merged in, I plan to add any missing M2M associations that haven't been implemented yet or anything else that may benefit from this new framework (e.g., #370, #624, #716).

[jvanderaa]

Is this something from firewall models? I'm not quite following this.

[joewesch]

Apparently these have been here since v2.0.0 (nautobot/nautobot#3399): https://github.com/nautobot/nautobot/blob/develop/nautobot/ipam/models.py#L1626-L1632

[glennmatthews]

These exist in the data model but aren't really fully baked at this time (i.e. they don't actually do anything) and aren't (IIRC) exposed in the UI either.

[joewesch]

Ah, ok I was unaware they weren't really being used. I was mostly going based on the serializer and showing up in the API Schema:

I can remove them.

[glennmatthews]

I don't think it hurts to include them, my comment was more of just an FYI.

[joewesch]

Note for reviewers: if you only have time to thoroughly review one file it should be this file. This is where almost all of the new logic lives. The rest of the files are primarily the field definitions and tests.

[gsnider2195]

I didn't see any problems in the module utils. I admit that I didn't understand some of the field name mappings.