Fix zombie ctags processes by adding waitFor() after destroyForcibly()

When ctags processing times out or is terminated, destroyForcibly() sends
SIGKILL but the parent never calls waitFor() to reap the child process.
This causes zombie processes to accumulate over long indexing runs.

Extract destroyProcess() helper that calls destroyForcibly() followed by
waitFor(10, SECONDS) to properly reap terminated ctags processes.

Fixes #4920

Author: gaborbernat

opengrok-indexer/src/main/java/org/opengrok/indexer/analysis/Ctags.java

@@ -142,7 +142,19 @@ public void close() {
         if (ctagsProcess != null) {
             closing = true;
             LOGGER.log(Level.FINE, "Destroying ctags command");
-            ctagsProcess.destroyForcibly();
+            destroyProcess();
+        }
+    }
+
+    private void destroyProcess() {
+        ctagsProcess.destroyForcibly();
+        try {
+            if (!ctagsProcess.waitFor(10, TimeUnit.SECONDS)) {
+                LOGGER.log(Level.WARNING, "ctags process did not terminate within timeout");
+            }
+        } catch (InterruptedException e) {
+            LOGGER.log(Level.WARNING, "Interrupted while waiting for ctags process to terminate");
+            Thread.currentThread().interrupt();
         }
     }
 
@@ -603,10 +615,10 @@ private void readTags(CtagsReader reader) throws InterruptedException {
                         }
                     } catch (IllegalThreadStateException e) {
                         LOGGER.log(Level.WARNING, "ctags EOF but did not exit");
-                        ctagsProcess.destroyForcibly();
+                        destroyProcess();
                     } catch (Exception e) {
                         LOGGER.log(Level.WARNING, "ctags problem:", e);
-                        ctagsProcess.destroyForcibly();
+                        destroyProcess();
                     }
                     // Throw the following to indicate non-I/O error for retry.
                     throw new InterruptedException("tagLine == null");