Output of interpolated time series should not include the columns used to derive the result series name

[jvshahid]

For example for this query select count(c1) from points group by time(5s), c2 into :series_name.count.[c2]; and data:

[
    { "name": "points",
      "columns": ["c1", "c2", "time"],
      "points": [
        [1, "aa", 1398893416],
        [2, "aa", 1398893416],
        [3, "aa", 1398893416],
        [7, "bb", 1398893416],
        [8, "bb", 1398893416],
        [9, "bb", 1398893416]
    ]}
  ]

The resulting time series shouldn't have the c2 column, since that is implied from the series name. It's more efficient to not include the same value over and over in the output time series.