fix(gooddata-sdk): resolve ty + dbt + htmltest CI failures

Three small fixes uncovered by CI on the previous push:

- `dataset.py` (gooddata-sdk): the new `type: Literal["NORMAL",
  "AUXILIARY"] | None` field on `CatalogDeclarativeDataset` shadowed the
  builtin `type` inside the class body, so `type[DeclarativeDataset]` in
  `client_class()` resolved to the field (None) and tripped ty with
  `invalid-type-form`. Use `builtins.type[...]` (the same pattern
  `setting.py` and `identifier.py` already use) and import `builtins`.
- `metrics.py` (gooddata-dbt): `attribute.source_column`,
  `label.source_column`, and `fact.source_column` are now `str | None`
  on the SDK side (AUXILIARY datasets carry synthetic entries with no
  physical column). Guard the `.lower()` calls so AUX entries are
  silently skipped instead of crashing.
- `get_hll_type.md` and `ai-lake/analyze_statistics.md`: relative links
  to sibling pages were missing trailing slashes, which htmltest rejects
  ("target is a directory, href lacks trailing slash"). Added them.

risk: low

Author: jaceksan

docs/content/en/latest/administration/organization/get_hll_type.md

@@ -15,7 +15,7 @@ Reads the organization-level `hyperLogLogType` setting. Returns the
 configured value (`"Native"` or `"Presto"`), or `None` when the setting
 is unset or carries an unrecognized value.
 
-See [`set_hll_type`](../set_hll_type) for the meaning of each value and
+See [`set_hll_type`](../set_hll_type/) for the meaning of each value and
 when to choose `"Native"` versus `"Presto"`.
 
 {{% parameters-block title="Returns"%}}

docs/content/en/latest/data/ai-lake/analyze_statistics.md

@@ -18,8 +18,8 @@ pre-aggregation table whose dimension attributes the platform will later
 resolve via filter pushdown.
 
 The call returns immediately with an operation ID; the actual analyze
-runs asynchronously. Use [`get_operation`](../get_operation) or
-[`wait_for_operation`](../wait_for_operation) to poll for completion.
+runs asynchronously. Use [`get_operation`](../get_operation/) or
+[`wait_for_operation`](../wait_for_operation/) to poll for completion.
 
 {{% parameters-block  title="Parameters"%}}
 

packages/gooddata-dbt/src/gooddata_dbt/dbt/metrics.py

@@ -86,14 +86,18 @@ def get_entity_type(self, table_name: str, expression_entity: str) -> str:
             if table_id is not None and table_id.id.lower() == table_name.lower():
                 attributes = dataset.attributes if dataset.attributes else []
                 facts = dataset.facts if dataset.facts else []
+                # `source_column` is optional on the SDK side — AUXILIARY
+                # datasets carry synthetic attributes/labels/facts with no
+                # physical column. Skip those entries here, since they have
+                # no source column to match against.
                 for attribute in attributes:
-                    if attribute.source_column.lower() == expression_entity_cmp:
+                    if attribute.source_column and attribute.source_column.lower() == expression_entity_cmp:
                         result = "label"
                     for label in attribute.labels:
-                        if label.source_column.lower() == expression_entity_cmp:
+                        if label.source_column and label.source_column.lower() == expression_entity_cmp:
                             result = "label"
                 for fact in facts:
-                    if fact.source_column.lower() == expression_entity_cmp:
+                    if fact.source_column and fact.source_column.lower() == expression_entity_cmp:
                         result = "fact"
         for date_dataset in self.ldm.ldm.date_instances:
             if date_dataset.id.lower() == expression_entity_cmp:

packages/gooddata-sdk/src/gooddata_sdk/catalog/workspace/declarative_model/workspace/logical_model/dataset/dataset.py

@@ -1,6 +1,7 @@
 # (C) 2022 GoodData Corporation
 from __future__ import annotations
 
+import builtins
 from pathlib import Path
 from typing import Literal
 
@@ -67,7 +68,10 @@ class CatalogDeclarativeDataset(Base):
     workspace_data_filter_references: list[CatalogDeclarativeWorkspaceDataFilterReferences] | None = None
 
     @staticmethod
-    def client_class() -> type[DeclarativeDataset]:
+    def client_class() -> builtins.type[DeclarativeDataset]:
+        # `builtins.type[...]` (not bare `type[...]`) because the class
+        # defines a `type: Literal[...]` field that shadows the builtin
+        # inside the class body, which trips ty.
         return DeclarativeDataset
 
     def store_to_disk(self, datasets_folder: Path, sort: bool = False) -> None: