write csv: line protocol malformed if there is a column named result

[russorat]

#datatype,string,long,dateTime:RFC3339,dateTime:RFC3339,dateTime:RFC3339,double,string,string,string,string,string,string,string,string,string,string
#group,true,false,false,false,false,false,true,true,true,true,true,true,true,true,true,true
#default,_result,,,,,,,,,,,,,,,
,result,table,_start,_stop,_time,_value,_field,_measurement,env,host,hostname,nodename,org,result,role,url
,,0,2020-08-26T23:10:54.023607624Z,2020-08-26T23:15:54.023607624Z,2020-08-26T23:11:00Z,0,0.001,something,host,pod,node,host,,success,role,http://127.0.0.1:8099/metrics

when run with the following command:

influx write dryrun -c default -b apps --format csv -f sample.csv
something,env=host,host=pod,hostname=node,nodename=host,result=_result,result=success,role=role,url=http://127.0.0.1:8099/metrics 0.001=0 1598483460000000000

The line protocol is not valid, because the result column is in there twice. removing the result column (tag) from the data solves the issue.

[sranka]

The flux query result is wrongly encoded, the first result column is not part of the group key and should have false in the #group annotation row. The same duplication would happen with a "table" tag. This is now reported in influxdata/flux#3161, a fix is proposed in influxdata/flux#3162.

As an enhancement, the csv2lp conversion should detect duplicate tag names, warn about them and use the last one.