Summarize Simplesat results without mixing metrics

**Objective:** Produce a summary of Simplesat survey activity between {start-date} and {end-date}, optionally limited to the survey {survey-name}.

**Step 1: Establish which surveys are in scope**

Call `simplesat-list-surveys`. If {survey-name} is supplied, pass it as the name filter. Record each survey's ID, Name and Metric. The metric tells you what the scores mean (for example CSAT, NPS or CES); use it when you describe the numbers, and never mix two metrics into one average.

**Step 2: Count the responses**

Call `simplesat-list-responses` with created_from set to {start-date} and created_to set to {end-date}. If {survey-name} is supplied, also pass survey_name. Count the responses per Survey Name and note the first and last Created dates actually seen.

**Step 3: Distribute the answers**

Call `simplesat-list-answers` with created_from set to {start-date} and created_to set to {end-date}, once per survey in scope using survey_name. Group the answers by Choice Label and by sentiment, and count each group.

**Step 4: Report**

For each survey, in this order:

* a heading with the survey name and its metric
* response count for the period
* the answer distribution as a table of Choice Label, count, and percentage of that survey's answers
* the split of positive, neutral and negative sentiment
* one sentence on what changed or stands out

End with a single overall line giving the total responses across all surveys in scope.

**Rules & Guidelines:**

* This prompt is read only. Do NOT call `simplesat-update-answer`, `simplesat-add-customer` or `simplesat-send-survey-email`.
* Percentages are of the answers returned, not of anything else. Say so in the table caption.
* Do not compute an average score across surveys with different metrics.

**Validation:**

* If {start-date} or {end-date} is missing or not a date, ask for both in YYYY-MM-DD form and stop.
* If {end-date} is before {start-date}, say so and stop.
* If no responses fall in the range, report zero for the period and state the dates used rather than silently widening them.