Merge bitcoin#31026: ci: set a ctest test timeout of 1200 (20 minutes)

56aad83 ci: set a ctest timeout of 1200 (20 minutes) (fanquake)

Pull request description:

  This should be long enough (with headroom) for our longest running tests, which even under MSAN, TSAN, Valgrind, etc max out at about 800s.

  i.e under Valgrind I see the longer runtimes as:
  ```bash
  135/136 Test   #8: bench_sanity_check_high_priority .....   Passed  371.19 sec
  136/136 Test bitcoin#122: coinselector_tests ...................   Passed  343.39 sec
  ```

  In the CI `tests` [under TSAN](https://cirrus-ci.com/task/6321297691508736?logs=ci#L2520):
  ```bash
  tests ................................   Passed  795.20 sec
  ```
  [and MSAN](https://cirrus-ci.com/task/4913922807955456?logs=ci#L2226):
  ```bash
  tests ................................   Passed  658.48 sec
  ```

  This will also prevent the current issue we are seeing of `ctest` running until it reaches the CI timeout, see bitcoin#30969.

  We still need to figure out what underlying issue is causing the tests to (sometimes) run for so long, but in the mean time, this will stop `ctest` wasting our CI CPU. It should also make it more clear in the logs, exactly which test is the one that is hitting the timeout.

ACKs for top commit:
  maflcko:
    review ACK 56aad83
  tdb3:
    re ACK 56aad83

Tree-SHA512: 43c0dc12b8b12b1d9804751a9816935e2abbe962b451e12a268f2d2c430bc568b83995dbc405f100b596dfb0f1e9f65b78074de98916592d3ae4ebc2126e3a6c

Author: fanquake

ci/test/03_test_script.sh

@@ -146,7 +146,7 @@ if [ "$RUN_CHECK_DEPS" = "true" ]; then
 fi
 
 if [ "$RUN_UNIT_TESTS" = "true" ]; then
-  DIR_UNIT_TEST_DATA="${DIR_UNIT_TEST_DATA}" LD_LIBRARY_PATH="${DEPENDS_DIR}/${HOST}/lib" CTEST_OUTPUT_ON_FAILURE=ON ctest "${MAKEJOBS}"
+  DIR_UNIT_TEST_DATA="${DIR_UNIT_TEST_DATA}" LD_LIBRARY_PATH="${DEPENDS_DIR}/${HOST}/lib" CTEST_OUTPUT_ON_FAILURE=ON ctest "${MAKEJOBS}" --timeout $((TEST_RUNNER_TIMEOUT_FACTOR * 30 ))
 fi
 
 if [ "$RUN_UNIT_TESTS_SEQUENTIAL" = "true" ]; then