Bump deprecations

[teunbrand]

This PR does not fix an issue, but I see it as regular maintenance.

Briefly, it converts deprecate_warn0() to lifecycle::deprecate_stop() for functions/arguments that were deprecated prior to ggplot 3.3.0.
It converts deprecate_soft0() to deprecate_warn0() for functions/arguments that were deprecated up to ggplot2 3.2.0 and removes examples.

I thought 5 years to be an OK horizon for deprecation warnings and I felt it was time to stop allowing things from the previous major version. Happy to make these timelines shorter though :)

[thomasp85]

Let's have a revdepcheck for this before we fully commit. I agree in principle about the timespan

[thomasp85]

revdepchecks have been committed

[teunbrand]

So about 30/347 packages show up with at least a deprecation warning. This comes with the caveat that we only see warnings when something else is wrong too: I don't think throwing a warning includes a package in the problems section.

I think we might be slightly too aggressive about erroring on the theme settings, as giving a wrong argument usually gives a warning, not an error. Honestly I'm kind of baffled that some packages still use arguments that were deprecated since ggplot2 2.0.0 (i.e. 10 years ago).

[teunbrand]

Per #6281 (comment):

I think we are at the point were we can start a deprecation process. We already have deprecation warnings for default_aes to alert extension developers and 2 years in I think we can begin to make noise for the users

Regarding size to linewidth translation