Fix documentation build errors

The documentation pages for the Pydantic types are now generated but
not included in the table of contents, so they have to be exxluded
for now in order to avoid generating orphaned pages. Also update the
nitpick exclusions to remove some that are now unnecessary and add
another case where autodoc_pydantic generates a spurious link.

docs/api.rst

@@ -72,8 +72,14 @@ API reference
 .. automodapi:: safir.middleware.x_forwarded
    :include-all-objects:
 
+..
+   If all objects from safir.pydantic are included, the pages for the
+   Pydantic types are generated but not linked into any table of contents.
+   This is probably a but in automodapi and should be checked against later
+   verisons to see if it's fixed. Once this is fixed, the nitpick exclusions
+   in documenteer.yaml can also be removed.
+
 .. automodapi:: safir.pydantic
-   :include-all-objects:
 
 .. automodapi:: safir.redis
    :include-all-objects:

docs/documenteer.toml

@@ -17,19 +17,20 @@ nitpick_ignore_regex = [
     ['py:.*', 'starlette.*'],
 ]
 nitpick_ignore = [
-    ['py:class', 'unittest.mock.Base'],
-    ['py:class', 'unittest.mock.CallableMixin'],
+    # These don't appear to have documentation but show up in the inheritance
+    # tree.
+    ["py:class", "unittest.mock.Base"],
+    ["py:class", "unittest.mock.CallableMixin"],
+    # autodoc_pydantic generates some spurious links that can't be resolved.
     ["py:obj", "ComputedFieldInfo"],
-    ["py:obj", "JobMetadata.id"],
-    ["py:class", "pydantic.BaseModel"],
-    ["py:class", "BaseModel"],
+    ["py:class", "lambda"],
     # arq doesn't provide documentation for all of its types.
     ["py:class", "arq.cron.CronJob"],
     ["py:class", "arq.typing.StartupShutdown"],
     ["py:class", "arq.typing.WorkerCoroutine"],
     ["py:class", "arq.worker.Function"],
-    # sphinx-automodapi apparently doesn't recognize TypeAlias as an object
-    # that should have generated documentation, even with include-all-objects.
+    # Including the documentation for these generates the pages but doesn't
+    # include them in the table of contents, so for now they're excluded.
     ["py:obj", "safir.pydantic.EnvAsyncPostgresDsn"],
     ["py:obj", "safir.pydantic.EnvRedisDsn"],
     ["py:obj", "safir.pydantic.HumanTimedelta"],