Is it appropriate to put a scientific/technical document about code developed by our group on arXiv?

If your manual is specific to your group's environment (software, lab, equipment, protocols, etc.), then I would consider posting such document on arXiv as not appropriate. A better place for such documents IMHO would be figshare or Zenodo. Both services allow artifacts to be not only citable, but also discoverable (via DOI assignment). Version control is supported as well, but the advantage of Zenodo in this regard in comparison with figshare is Git and GitHub integration.

If, on the other hand, the document is not specific to your group's or other particular environment (or, perhaps, you want to share this document as a model of how such documents should be structured, presented and what content should they contain), then I agree with other answers in that it is totally appropriate to post such document on arXiv.

Yes, this seems like a totally appropriate use of the arXiv. If you hope that other people will use your code, then the documentation has to be available online. While you will presumably have the documentation available alongside the code (wherever you are making the latter available), putting the documentation on the arXiv as well will certainly make more people aware of the tools you have created.

As you say, the documentation for some scientific analysis programs are already available on the arXiv. These files form a very small subset of the total arXiv, but they are valuable. Once or twice I have looked at the documentation for analysis programs on the arXiv, even though I was not using (or even considering using) the software packages myself; I looked up the documentation because it helped me understand other people's papers that did use these tools.