Compress API results before caching them

Some result sets may be larger than the default cache's 1 MB max item size. This change is a bandaid meant to allow us to resume use of memcached while we work on a longer-term fix.

LEARNER-682

openedx/core/lib/edx_api_utils.py

@@ -8,6 +8,7 @@ from django.core.exceptions import ImproperlyConfigured
 from edx_rest_api_client.client import EdxRestApiClient
 from provider.oauth2.models import Client
 
+from openedx.core.lib.cache_utils import zpickle, zunpickle
 from openedx.core.lib.token_utils import JwtBuilder
 
 
@@ -47,10 +48,11 @@ def get_edx_api_data(api_config, user, resource, api=None, resource_id=None,
 
     if cache_key:
         cache_key = '{}.{}'.format(cache_key, resource_id) if resource_id is not None else cache_key
+        cache_key += '.zpickled'
 
         cached = cache.get(cache_key)
         if cached:
-            return cached
+            return zunpickle(cached)
 
     try:
         if not api:
@@ -89,7 +91,8 @@ def get_edx_api_data(api_config, user, resource, api=None, resource_id=None,
         return no_data
 
     if cache_key:
-        cache.set(cache_key, results, api_config.cache_ttl)
+        zdata = zpickle(results)
+        cache.set(cache_key, zdata, api_config.cache_ttl)
 
     return results