Fix unintended control flow in fellow_cache_lru_work()

The optimized case for multiple segments from the same fco did
not work as expected, continue did not continue the inner loop.

Spotted by Coverity, CID#469236

src/fellow_cache.c

@@ -3182,7 +3182,7 @@ fellow_cache_lru_work(struct worker *wrk, struct fellow_cache_lru *lru)
 			continue;
 		}
 		if (fcs->state == FCS_INCORE) {
-			do {
+			while (1) {
 				fellow_cache_lru_seg_evict_locked(fcs,
 				    &alloc, lru);
 				AN(buddy_return_ptr_extent(rets, &alloc));
@@ -3195,7 +3195,8 @@ fellow_cache_lru_work(struct worker *wrk, struct fellow_cache_lru *lru)
 					fcss = VTAILQ_NEXT(fcs, lru_list);
 					continue;
 				}
-			} while (0);
+				break;
+			}
 
 			AZ(pthread_mutex_unlock(&fco->mtx));
 			fco = NULL;