fix(types): make schema.plugin() more flexible for schemas that don't define any generics

[vkarpov15]

Fix #12454
Re: #12139

Summary

I think just switching to any should be enough to fix #12454, it doesn't break any of the tests from #12139. Any thoughts @emiljanitzek ?

Examples

[emiljanitzek]

This would make the plugin function type a lot more loose and you would not get the same type safety. So I would prefer it not to change. I would say that the problem is that the Schema class is too strictly typed by default. I haven't tested this but I think changing the Schema class generics should have the same effect without changing the plugin interface.

export class Schema<
    EnforcedDocType = any,
    M = Model<EnforcedDocType, any, any, any>,
    TInstanceMethods = any,
    TQueryHelpers = any,
    TVirtuals = any,
    TStaticMethods = any,
    TPathTypeKey extends TypeKeyBaseType = DefaultTypeKey,
    DocType extends ObtainDocumentType<DocType, EnforcedDocType, TPathTypeKey> = ObtainDocumentType<any, EnforcedDocType, TPathTypeKey>>
    extends events.EventEmitter {

[hasezoey]

LGTM, somehow i missed this as a possible solution when making #12458

[vkarpov15]

The unfortunate issue is that changing TInstanceMethods = {} to TInstanceMethods = any causes a lot more problems with schema type inference, and those problems will be harder to fix. The suggestion to make schema generics default to any is an interesting one, we'll have to consider it further. But, for now, we should loosen the types up a bit to avoid potentially breaking changes.