White space characters between XML tags (used for formatting purposes only) are outputted into FI stream.

The outputted FI stream/file contains whitespaces which were placed between tags in original XML and used to create a "pretty" XML (not single line -> added line breaks and indetion to make it easier to read for a human). Those whitespaces are not removed from the stream and are passed into the FI stream making it unnecesarilly larger.
This issue was discovered while I was comparing multiple Fast Infoset implementations - there is a noticable size difference and the additional whitespaces are clerarly identifiable in the stream while using a HEX editor (PSPad, in my case).
I can provide test data if necessary.
#### Environment
MS Windows 7 x64 - Oracle Java 7 update 40 (x64)
Fedora 18 x64 - OpenJDK 7 update 45
#### Affected Versions
[1.2.9]