Add lookup() / related() high-level helpers and fresh() cache scope

Two new wrappers fold the common four-step pattern (resolve → relations
→ entities → join) into single calls. Both accept friendly id_type
aliases (``name``, ``chebi``, ``hmdb``, ``uniprot``, …) and pivot the
matching identifiers into named columns; ``related()`` does the
subject + object join and exposes filters by sources, predicates,
relation_categories, and participant_types (also aliased:
``protein``, ``small_molecule``, …). Multi-valued aliases pick the
shortest representative.

Also adds ``cache_clear()`` and a ``fresh()`` context manager: any
URL+payload first-touched inside the block is re-downloaded; later
identical requests in the same scope hit the freshly populated cache.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: deeenes

omnipath_client/__init__.py

@@ -33,20 +33,28 @@
     'endpoints',
     'params',
     'values',
+    'lookup',
+    'related',
+    'cache_clear',
+    'fresh',
 ]
 
 from omnipath_client import cosmos, utils
 from ._client import (
     OmniPath,
+    fresh,
+    lookup,
     params,
     values,
+    related,
     entities,
     resolve,
     endpoints,
     relations,
     resources,
     annotations,
     ontologies,
+    cache_clear,
     search_terms,
     ontology_tree,
     ontology_terms,