WIP: use copier

[thomaspaulb]

• When starting a new Waft project, use copier instead of git clone
• When updating an existing Waft project, use copier instead of editing and re-running bootstrap
• When updating, don't override local README.md
• Check how to make copier pick as a default the answers that were chosen last time.
• When turning an existing Waft project into a copier project by using "copier copy" over the thing, guess default answers to questions based on the files that are already there, using context hooks.
• Test template versioning. How does it work? with git tags? Yes with tags, it looks for the latest tag. Also in other branches. Tested with github/thomaspaulb/test-template
• Test the use of migration scripts during copier update
• Try to use migration scripts to replace symlinks with actual copies, and if that is desirable.
• Satisfy Hussam's requirement that it should be possible to be on, for example, "waft v3" and still receive updates, while other projects are already on "waft v4". This can be done by implementing https://github.com/copier-org/copier/issues/507, while having the template specify that major versions have breaking changes and so an update across major versions should be intentional; a regular ./bootstrap would just use copier update and update only minor versions.
• If bootstrap runs copier update, we still need to do other "bootstrap-like" things afterwards eg checkout a new version of waftlib and run commands there. Have to think about whether this should be a task that is always triggered after copier update, or a migration script, or extra lines in the bootstrap script.
• Try to add for example a tool dependency in pyproject.toml, and see how that works on a project upgrade.