Add support for ignoring trailing line(s) in CSV

## Background
Sometimes input raw files contain a trailing line that needs to be ignored. We have a use case for the trailing 1 line, but if we need to implement this feature anyway, maybe let's consider do it more generally.

## Feature
Add support for a command-line option to ignore trailing lines of each input CSV file.

## Example [Optional]
```
a,b,c
1,2,3
5,6,7
Total=2
```

## Proposed Solution [Optional]
If there is no such option in `spark-csv`, possibly we need to preprocess incoming CSV files in RDDs.