nbdev1 migration scripts and documentation needs to be improved

I am doing a migration from nbdev1 to nbdev2 for my project, following https://nbdev.fast.ai/migrating.html document, and I have encountered the following problems.

- The nbdev_migrate script has some problems where the page description (which in nbdev1 was done as a quote just below h1 header) spanned more than one line; only the first line was copied.
- The nbdev_migrate script automatically converts directives, but it does not convert the old `notebook2script` calls
- In _"Update directive names"_ there is no need to do the convert directives, as it is handled by nbdev_migrate script
- With the exception of `git add $LIBNAME/_modidx.py` the migration documentation does not say anything about adding (or not) those newly downloaded files to the repository with `git add` - should it be done (and why), should it not be done (and why)?
- Instruction on how to update GitHub Pages settings is done **3** times.
- I think instead of using perl one can simply use sed in those examples (though I am not exactly sure which is easier for an ordinary user)

Start with migrating.html and the nbdev_migrate script, checking the listed migration steps, directive handling, multiline page descriptions, and old notebook2script calls. Review the guidance around git add, GitHub Pages settings, and perl versus sed. Done means the script and migration documentation consistently cover each reported case without duplicated instructions.

Written by the indexing model from the issue text.