Document Kapacitor-idiomatic way for running advanced-syntax InfluxQL functions

There are known issues (https://github.com/influxdata/kapacitor/issues/1978) when trying to execute a batch query written in InfluxQL that uses functions to compute the `difference`, `derivative`, etc. on an aggregate.

For example:

```
SELECT DERIVATIVE(MEAN("water_level"))
FROM "h2o_feet"
WHERE "location" = 'santa_monica'
AND time >= '2015-08-18T00:00:00Z'
AND time <= '2015-08-18T00:30:00Z'
GROUP BY time(12m)
```

As the [documentation](https://docs.influxdata.com/kapacitor/v1.5/nodes/query_node/) says, you cannot put any `group by` or `where time...` in the `QueryNode`, because they will be overwritten by Kapacitor when you add `.groupBy(...).period(...)...`. However, the advanced syntax for those functions __requires__ `group by time(...)`.

I propose that the documentation explicits an example to compute the `non_negative_difference` (one for all those functions) in Kapacitor-idiomatic way.

The solution is to let Kapacitor compute the non negative difference outside of the query node:

```
batch
|query('''
SELECT mean("value")
FROM "telegraf"."default".cpu_usage_idle
WHERE "host" = 'serverA'
''')
.period(1m)
.every(20s)
.groupBy(time(10s), 'cpu')
| difference('max_usage')
| where(lambda: "difference" >= 0)
```

Start with the linked Kapacitor QueryNode documentation and compare it with the issue's batch-query example. Document the Kapacitor-idiomatic approach for non_negative_difference, including how to use the difference and where nodes outside the query; done means a newcomer can follow the example without relying on advanced InfluxQL syntax.

Written by the indexing model from the issue text.