`ProjectSnapshot` - model, orchestrator, tests

[ravi-kumar-pilla]

Description

Implement the ProjectSnapshot dataclass and _build_project_snapshot() orchestrator that assembles all sub-snapshots into a single complete view of the project. This is the internal composition layer consumed by get_project_snapshot() in #5432.

Context

The following builders from prior PRs feed into this orchestrator:

• _build_project_metadata_snapshot(metadata) → ProjectMetadataSnapshot (ProjectMetadataSnapshot : model, builder, tests #5428)
• _build_pipeline_snapshots() → list[PipelineSnapshot] (NodeSnapshot + PipelineSnapshot - models, builder, tests #5429)
• _build_dataset_snapshots(config_loader) → dict[str, DatasetSnapshot] (DatasetSnapshot - model, builder, tests #5430)

Two functions deferred from earlier PRs as premature now belong here:

• _make_config_loader(project_path, env) - removed from DatasetSnapshot - model, builder, tests #5430; lives here as the orchestrator creates the loader once and passes it to both _build_dataset_snapshots and _build_parameter_keys
• _get_parameter_keys(config_loader) - removed from DatasetSnapshot - model, builder, tests #5430; parameters are a list[str] of keys with no values exposed, a separate concern from DatasetSnapshot

Design decisions:

• bootstrap_project() not _get_project_metadata() - bootstrap_project() additionally calls configure_project() which populates the global pipelines mapping required by _build_pipeline_snapshots()
• Config loader created once — _make_config_loader uses settings.CONFIG_LOADER_CLASS (not hardcoded OmegaConfigLoader) and settings.CONFIG_LOADER_ARGS; the single instance is passed to both _build_dataset_snapshots and _build_parameter_keys
• _resolve_factory_patterns uses CatalogConfigResolver.resolve_pattern() only — this resolves template variables ({namespace}, {name}) for pipeline-referenced datasets that match a factory pattern but have no explicit catalog entry. It does NOT use credential resolution, so it is safe in a credentials-free context. Input datasets dict is not mutated.
• Parameters are list[str] — keys only ("parameters", "params:model_options", "params:model_options.test_size"), never values. Nested keys are recursively flattened and sorted.

Scope & Deliverables:

• kedro/inspection/models.py - Add ProjectSnapshot:

@dataclass
class ProjectSnapshot:
    metadata: ProjectMetadataSnapshot
    pipelines: list[PipelineSnapshot]
    datasets: dict[str, DatasetSnapshot]
    parameters: list[str] = field(default_factory=list)

• kedro/inspection/snapshot.py - Add:

def _build_project_snapshot(project_path: Path, env: str) -> ProjectSnapshot:
    # orchestrator: bootstrap_project → config_loader → all builders → ProjectSnapshot

• kedro/inspection/helper.py - Add:

Move any helpers that are not snapshot builders here

def _make_config_loader(project_path: Path, env: str) -> AbstractConfigLoader:
    # uses settings.CONFIG_LOADER_CLASS and settings.CONFIG_LOADER_ARGS

def _get_parameter_keys(config_loader: AbstractConfigLoader) -> list[str]:
    # sorted list of "parameters" + all "params:..." keys, no values

def _resolve_factory_patterns(
    datasets: dict[str, DatasetSnapshot],
    conf_catalog: dict[str, Any],
    pipelines: list[PipelineSnapshot],
) -> dict[str, DatasetSnapshot]:
    # enriches datasets with pipeline-referenced names that match factory patterns
    # uses CatalogConfigResolver.resolve_pattern() only, no credential resolution
    # does not mutate input datasets dict

• tests/inspection/test_project_snapshot.py