Know whether a Ghost Inspector suite is getting better or worse

**Objective:** Report the health of the Ghost Inspector suite named `{suite-name}` over the last `{lookback-days}` days (default 30).

**Step 1: Identify the suite**

Call `ghost-inspector-get-suite` with `name.contains` set to `{suite-name}`. If more than one matches, list them with their organization and folder and stop. Keep the suite `id`, `test-count`, `browser`, `region` and `test-frequency`.

**Step 2: Pull the suite run history**

Call `ghost-inspector-get-suite-result` with `suite-id` from step 1 and `date-created.from` set to the start of the window. For each run you have `count-passing`, `count-failing`, `count-unknown`, `execution-time-ms` and `date-execution-started`.

**Step 3: Work out the trend**

* Overall pass rate across the window, and pass rate for the most recent quarter of it, so a recent decline is visible
* Median run duration for the first half of the window against the second half; call out a duration increase above 25 percent
* Any run with a non zero `count-unknown`, which means tests neither passed nor failed and usually points at an infrastructure problem

**Step 4: Find the tests responsible**

Call `ghost-inspector-get-test-result` with `suite-id` from step 1, `passing` false and the same `date-created.from`. Group by `test-name` and count. These are the tests costing the suite its pass rate.

**Step 5: Report**

* Suite name, organization, folder, test count, schedule, default browser and region
* Pass rate over the window, and whether it is improving or declining
* Median and worst run duration, and the trend
* A table of the failing tests by failure count
* Two or three concrete recommendations

**Rules and guidelines:**

* Read only. Do NOT call `ghost-inspector-execute-suite`, `ghost-inspector-execute-test`, `ghost-inspector-update-suite`, `ghost-inspector-duplicate-suite`, `ghost-inspector-cancel-suite-run` or any comment tool.
* Report the numbers you actually have. Do not extrapolate a trend from fewer than four runs; say the sample is too small instead.

**Validation:**

* If `{suite-name}` matches nothing, list some suite names from `ghost-inspector-get-suite` and stop.
* If the suite has no results in the window, say it has not run in that period and check its `test-frequency` to say whether that is expected.