Add page with instructions for installing old versions of napari

[jni]

📚 New content request

Python's dependency management is not ideal for installing old versions of libraries, because metadata in PyPI is static and upper version constraints are generally not a good idea for libraries. This means that installing old versions of napari for whatever reason (reproducing a paper or using a library that is no longer being updated) is very hard.

Since we started using version constraints/pinning for CI, from version 0.4.18 we have a record of all the library dependencies, both direct and indirect, that work for a specific napari release. (See this comment.) This is a great and unexpected* version of the pinning! (*at least by me 😅)

Additionally, I recently came across PyPI time machine, which lets you install packages with pip while pretending to live in the past.

It would be good to have a page, or a section in the installation guide, that points readers to these resources (including direct links to the constraints files).

Potentially, we should also generate equivalent conda environment definitions for each release.