Trait bounds for generic compact types

[dvdplm]

If a type param is used in a compact field, don't add TypeInfo + 'static bound (just 'static)
If a type param is used in a compact field, call meta_type::<<$ty as HasCompact>::Type>() instead of meta_type::<$ty>() in the implementation of TypeInfo.

Avoid adding duplicate bounds.

[dvdplm]

@ascjones CI is angry about funty. Would be good to make up our minds about this horrible fix.

[ascjones]

In the absence of better solutions, let's just add that fix to the CI script with a small PR into master so we don't need to do it in each PR. We can remove it if/when we update parity-scale-codec to a version which fixes the issue.

[Robbepop]

arg my review never got sent ...

[Robbepop]

How about also implement manual custom trait bounds for derived code to resolve this problem?

[dvdplm]

How about also implement manual custom trait bounds for derived code to resolve this problem?

I think we should make the derive as easy to use as possible. Adding these bounds manually requires users to have a firm grasp of both this crate and parity-scale-info's Compact machinery, which imo is a bit unfortunate. Adding manual trait bounds should be a "last resort" option and normal usage should not require it. This is just a personal opinion and it could very well be that the line should be drawn somewhere else as to what is "too advanced" and what isn't.

[Robbepop]

I think this can be closed in favor of our custom trait bounds implementation, right?

[ascjones]

I think this can be closed in favor of our custom trait bounds implementation, right?

It looks like it, @dvdplm can close if he agrees.

[dvdplm]

Closing as stale.