Why load a document only to have it mangled by converting it to the internal format of some online text editor?
When loading a document, any document, that you want to edit and then save back, there should be no conversion whatsoever. The question of how good support for ODF is, should not be 'how badly does it mangle my documents?'. It should be a given that the document is *not* mangled. The question on how good the support for ODF, or any file format, is, should be: 'what types of edits can this program make on this file format.'
For decades, we're accepting that documents editors save back a file that, on the binary level, is almost, but not quite, entirely unlike the original file. How weird this leniency towards document editors is, becomes apparent when looking at at how computer programmers work with documents. Computer programmers always use plain text files for everything. When the text editors they use saves their documents with tabs instead of spaces, or utf16 instead of utf8, they get quite irate and will abandon that text editor forever. Why do normal users not get angry at document editors that mangle their documents?
So instead of choosing these horrible black box online text editors, I advise you to use something like WebODF. This ODF editor, which is purely client-side javascript, can run on your private site and saves your ODF back as it found it with changes only in the places where you edited the document.