Add required start and end nodes and allow chatting to subset of pipelines

[proteusvacuum]

Description

This adds the requirement that there is always a single Start and End node, and that there is an accessible path between them.

When building up the pipeline graph, we can now traverse from the start node to the end node, so testing subgraphs of the pipeline in the pipeline builder is now possible by moving where the start and end points to.

This also adds automatically incrementing "New Pipeline" names, and adds a migration to create start and end nodes for existing pipelines.

I also did a refactor of the pipeline tests to make them less verbose and make things a bit easier to read.

User Impact

Users will be required to have start and end nodes in their pipelines which clarifies the flow of data.

Demo

Screencast.from.2024-11-28.13-18-49.mp4

Fixes: #638

Docs

[codecov-commenter]

❌ 1 Tests Failed:

Tests completed	Failed	Passed	Skipped
785	1	784	0

View the top 1 failed tests by shortest run time

apps/service_providers/tests/test_models.py::TestServiceProviderModel::test_cannot_delete_provider_models_with_associated_pipeline

Stack Traces | 1.14s run time

.../service_providers/tests/test_models.py:81: in test_cannot_delete_provider_models_with_associated_pipeline
    provider_model = LlmProviderModel.objects.get(id=node.params["llm_provider_model_id"])
.../hostedtoolcache/Python/3.11.10....../x64/lib/python3.11.../db/models/manager.py:87: in manager_method
    return getattr(self.get_queryset(), name)(*args, **kwargs)
.../hostedtoolcache/Python/3.11.10....../x64/lib/python3.11.../db/models/query.py:649: in get
    raise self.model.DoesNotExist(
E   apps.service_providers.models.LlmProviderModel.DoesNotExist: LlmProviderModel matching query does not exist.

To view more test analytics, go to the Test Analytics Dashboard
📢 Thoughts on this report? Let us know!