server: fix bug related to job-size matching

Problem: a workunit could error out with unsent results.
The feeder skips such results, but the size_regulator counts them
and doesn't so doesn't promote any new results.
Solution: the feeder scans for results even with workunit errors.
If marks these results as state OVER, outcome DIDNT_NEED

db/boinc_db.cpp

@@ -1868,7 +1868,6 @@ int DB_WORK_ITEM::enumerate(
             " and r1.workunitid=workunit.id "
             " and workunit.appid=app.id "
             " and app.deprecated=0 "
-            " and workunit.error_mask=0 "
             " and workunit.transitioner_flags=0 "
             " %s "
             " %s "

sched/feeder.cpp

@@ -300,6 +300,24 @@ static bool get_job_from_db(
                 continue;
             }
             
+            // if the WU had an error, mark result as DIDNT_NEED
+            //
+            if (wi.wu.error_mask) {
+                char buf[256];
+                DB_RESULT result;
+                result.id = wi.res_id;
+                sprintf(buf, "server_state=%d, outcome=%d",
+                    RESULT_SERVER_STATE_OVER,
+                    RESULT_OUTCOME_DIDNT_NEED
+                );
+                result.update_field(buf);
+                log_messages.printf(MSG_NORMAL,
+                    "[RESULT#%u] WU had error, marking as DIDNT_NEED\n",
+                    wi.res_id
+                );
+                continue;
+            }
+
             // Check for collision (i.e. this result already is in the array)
             //
             collision = false;