MINOR: Moving the rollback out of lock in share partition (#20153)

Moving rollback out of lock, if persister returns a completed future for
write state then same data-plane-request-handler thread should not call
purgatory safeTryAndComplete while holding SharePartition's write lock.

Reviewers: Andrew Schofield <aschofield@confluent.io>, Abhinav Dixit
 <adixit@confluent.io>

core/src/main/java/kafka/server/share/SharePartition.java

@@ -958,14 +958,12 @@ public class SharePartition {
                     break;
                 }
             }
-
-            // If the acknowledgement is successful then persist state, complete the state transition
-            // and update the cached state for start offset. Else rollback the state transition.
-            rollbackOrProcessStateUpdates(future, throwable, updatedStates, stateBatches);
         } finally {
             lock.writeLock().unlock();
         }
-
+        // If the acknowledgement is successful then persist state, complete the state transition
+        // and update the cached state for start offset. Else rollback the state transition.
+        rollbackOrProcessStateUpdates(future, throwable, updatedStates, stateBatches);
         return future;
     }
 
@@ -1014,13 +1012,12 @@ public class SharePartition {
                     break;
                 }
             }
-
-            // If the release acquired records is successful then persist state, complete the state transition
-            // and update the cached state for start offset. Else rollback the state transition.
-            rollbackOrProcessStateUpdates(future, throwable, updatedStates, stateBatches);
         } finally {
             lock.writeLock().unlock();
         }
+        // If the release acquired records is successful then persist state, complete the state transition
+        // and update the cached state for start offset. Else rollback the state transition.
+        rollbackOrProcessStateUpdates(future, throwable, updatedStates, stateBatches);
         return future;
     }