Improvements in `prediction.svyglm`

[tzoltak]

trafficstars

• [x] a possible bug
• [ ] a question about package functionality
• [x] a suggested code or documentation change, improvement to the code, or feature request

It will be better if:

1. prediction.svyglm can get survey design objects as data argument.
2. prediction.svyglm didn't fail if data contains NAs (it fails now because all observations with NAs are removed from results returned by survey::predict.svyglm what is a different behavior than this of predict.glm).