This notifies users if they might encounter [this bug](kubeflow/kubeflow#6136).  Can be removed when upstream is fixed

Profile CRD is removed and instead included in the kubeflow-profiles charm (see kubeflow-profiles add-crd commit for more description)

Other misc:
* separate build-deploy and relate-to-profile integration tests
* add check to set BlockedStatus when model.name!=kubeflow (limitation due to [upstream bug](kubeflow/kubeflow#6136))
* fix status reporting to avoid __init__ overwriting existing error Status' (see below).
* fix race condition(?) in relations by bumping ops to 1.2.0.  Accessing relation data from the relation data bag was (we think?) blocked by a race condition in the ops library.  Bumping to ops==1.2.0 fixes the issue

Status reporting:
Setting ActiveStatus() in init caused the accidental resetting of still valid statuses. For example:
* `config_change` hook triggers and is handled by `main()`.  Say it then hits an `OCIImageResourceError` in `main()`, setting an `ErrorStatus()`
* `update_status` hook triggers and has no handler.  It still runs `__init__()` which passes the `except NoVersionsListed/except NoCompatibleVersions` block, which sets `ActiveStatus`.  There is no additional handling on `update_status`, so it ends with the app in `ActiveStatus`

This masks any errors triggered by other hooks, making it harder for CI and humans to notice and fix errors (this also might happen instantly if an `update_status` is queued behind some other hook.