To the OP, use M-x customize-group RET backup
There you'll find options for keeping number of backups and where they store.
git is the tool to rely on for version tracking at any granularity you decide, keep committing and logging your changes and you can go back to any point with the builtin vc tool, for example, C-x v l, for logs, C-x v =, for changes, C-x v v, for next step, see the manual f1-r
Should I worry that the utf-8 file I start editing becomes a text-raw-unix file on save and git commit over tramp/scp as indicated by the initial character on the modeline changing from U to t?
The messages buffer mentions tramp encoding/decoding using
Saving file /scp:s:/ftp/pub/users/x/index.gmi... Tramp: Encoding local file ‘/tmp/tramp.x.gmi’ using ‘(lambda (beg end) (let ((coding-system-for-write 'binary) (coding-system-for-read 'binary)) (apply #'tramp-call-process-region '(tramp-file-name scp nil nil s nil /ftp/pub/users/x/index.gmi nil) beg end (car (split-string env GZIP= gzip)) t t nil (cdr (split-string env GZIP= gzip)))) (base64-encode-region (point-min) (point-max)))’...done Tramp: Decoding remote file ‘/scp:s:/ftp/pub/users/x/index.gmi’ using ‘(base64 -d -i | env GZIP= gzip -d >%s)’...done Wrote /scp:s:/ftp/pub/users/x/index.gmi
Have a look at the GNU Texinfo manual
https://www.gnu.org/software/texinfo/manual/texinfo/
at the moment I use GNU Emacs and Plan 9 User Space with that, oh, and Git.
There's an article on osnews saying content farms are writing for search engine optimization leading to bots training bots and the human touch in writing is becoming unsustainable financially. I was concerned recommending a search on the words "texinfo manual" might not get you to the source text.