Fix parsing when element is between buffers

[hanneskuettner]

This PR fixes an issue where the characters method in the XMLParser is called with element text that is cut in half due to the buffer size of the underlying parser.
We cannot stop looking at the value that we are interested in after we got the first value, but rather until we encounter the end tag for this element.

[faruktoptas]

Thank you for your contribution. Can you provide a Rss URL or a sample Xml please? I will create a unit test for this.

[hanneskuettner]

Sure thing!
Those are just two files. One file with the pubDate right at the 8096 byte buffer size, which fails on the original version, since it cuts the pubDate in half (stop after April).
In the other file all the fields fit in one buffer and are not cut.

test_fails_on_old.txt
test_succeeds_on_old.txt

[faruktoptas]

This works but when try to parse this xml:
onediorss.txt
First item title is "". But it works without your modifications.

[hanneskuettner]

I will look into that.
Can you provide me with more example files that you use for testing?

[faruktoptas]

I will add more samples. Now I am working on unit testing. It is almost done.

[faruktoptas]

I added unit tests to dev branch. You can go on working.