How to handle 2 tables returned, one valid, one with an error

running this query on a similar dataset usually returns multiple tables, one with data and the other with an error.
```
from(bucket: "telegraf")
|> range(start: -1m)
|> group()
```
Here is the raw data returned:
[2020-03-19-10-45_chronograf_data.csv.zip](https://github.com/influxdata/influxdb/files/4356121/2020-03-19-10-45_chronograf_data.csv.zip)

The UI displays the data from the first table, but gives no indication to the user that there was an error returned.

As a user, when my query returns some data but also an error, I would like to be notified that an error was returned so i don't assume the data returned is complete.

Reproduce the query against a similar dataset and inspect the attached raw CSV response to confirm that one table contains data while another contains an error. Trace how the UI handles multiple returned tables, then make completion visible when partial results include an error without hiding the valid data.

Written by the indexing model from the issue text.