Saving via db.Col().Save() is quite ineffecient as it requires
http connection per each document and cannot work well for millions
of documents. This issue is addressed with batch requests which
have packed http requests into multipart data.
Fixed infinite recursion in db.Col() when CreateCollection()
intermittently fails
Minor fixes of error recognition and some refactoring
Saving via db.Col().Save() is quite ineffecient as it requires http connection per each document and cannot work well for millions of documents. This issue is addressed with batch requests which have packed http requests into multipart data.