question: Should SCRAM_*_PLUS be feature-gated by registry_static?

[duesee]

While working on #32, I thought for a moment that I messed something up (see screenshot below). But it seems that the SCRAM_*_PLUS statics are not feature-gated by registry_static. I guess they should be?

[dequbed]

That is … weird. None of the mechanisms should ever be feature gated on registry_static, all that does is add a compile-time registry for mechanisms that allows other crates to add mechanisms to rsasl without forcing configuration hassle on users.

This might be a bug in rustdoc; I'm adding an attribute to the non-plus variant if registry_static is enabled, but I don't do that for the PLUS variant as I'm not solid on how to handle channel binding yet and thus the channel binding -PLUS variants are explicit opt-ins right now.

[duesee]

Should we tackle this as part of #32 and see if we can get rid of registry_static on these variants just by moving the annotations a bit? I feel there could be similar instances, but it's difficult for me to spot them. So, maybe it's better if you do a review first and see what else is off :-) And then we fix everything at once?