(csv) Feature to ignore empty trailing lines

[cowtowncoder]

(from FasterXML/jackson-dataformat-csv#145 by @robmoore)

We have a case where we're reading documents produced by a third-party which contains trailing empty lines. It appears that the reader attempts to create empty POJO entries for these lines rather than ignoring them.

A work around for this is to simply read each value individually rather than using readValues(). Nevertheless, for the streaming case it would be nice to be able to use readValues().

[cowtowncoder]

Looks like #15 might be more general solution here; and possibly easier to implement?

[cowtowncoder]

I think #15 is the superset, and until proven otherwise, assume it covers this too.