theta encoding wrong for text mark

[ondave]

When using a text mark with theta ordinal encoding, the values get wrapped back onto the start rather than distributed evenly around the circle.

Example:

{
  "data":{
    "values":[
      {"quadrant":"N"},
      {"quadrant":"E"},
      {"quadrant":"S"},
      {"quadrant":"W"}
    ]
  },
  "encoding": {
    "theta":{
          "field":"quadrant",
          "type":"ordinal",
          "sort":null
        },
    "text":{
          "field":"quadrant", 
        }
  },
  "layer": [
    {
      "mark": {
        "type": "text",
        "radius":50
      }
    }
  ]
}

However, this example for an arc mark give 4 equal segments as expected:

{
  "data":{
    "values":[
      {"quadrant":"N"},
      {"quadrant":"E"},
      {"quadrant":"S"},
      {"quadrant":"W"}
    ]
  },
  "encoding": {
    "theta":{
          "field":"quadrant",
          "type":"ordinal",
          "sort":null
        },
    "text":{
          "field":"quadrant", 
        }
  },
  "layer": [
    {
      "mark": {
        "type": "arc",
        "radius":50
      }
    }
  ]
}

[PBI-David]

The first generates a point scale and the second a band scale for some reason.