New strategy to share implementation between sync and Lwt

[mjambon]

Most modules are now shared via symlinks except for the Monad module. This makes the source code straightforward and still avoids the complexity of multifile functors. As a side effect, the testo and testo-lwt libraries are now independent of one another.

This is a preliminary to #57 which would otherwise require code duplication.

I need to check that this works in semgrep. I removed Testo.update_func in the process, which was a function needed to convert Lwt tests to synchronous tests. I'm not sure if it's really needed or what to do about it. Update: the solution I found is to not use Testo.update_func: https://github.com/semgrep/semgrep/pull/10147

PR checklist:

• [x] Purpose of the code is evident to future readers
• [x] Tests are included or a PR comment includes a reproducible test plan
• [x] Documentation is up-to-date
• [ ] A changelog entry was added to CHANGES.md for any user-facing change

Check out CONTRIBUTING.md for more details.

[mjambon]

@amchiclet

I don't have much clarity about how the symlinks are set up

I'll add some kind of readme alongside the symlinks since this is an unusual setup.