[SILGen] Fix the type of closure thunks that are passed const references

[Xazax-hun]

This PR is another attempt at landing #76903. The changes compared to the original PR:

• Instead of increasing the size of SILDeclRef, store the necessary type information in a side channel using withClosureTypeInfo.
• Rely on SGFContext to get the right ClangType
• Extend BridgingConversion with an AbstractionPattern to store the original clang type.
• The PR above introduced a crash during indexing system modules that references foreign types coming from modules imported as implementation only. These entities are implementation details so they do not need to be included during serialization. This PR adds a test and adds logic to exclude such clang types in the serialization process.

rdar://131321096&141786724

[Xazax-hun]

This is a slightly different approach to fix the same bits as #82486. The serialization related changes are the same and already got PR reviews for that in #82486.

[rjmccall]

I'm a little worried about the serialization changes. Are we relying on preserving Clang function types more than we were before?

[Xazax-hun]

Are we relying on preserving Clang function types more than we were before?

Yes, but we need this information to properly do type lowering for closure arguments, so I do not see a way around this without a major redesign. What are your concerns?

[rjmccall]

Are we relying on preserving Clang function types more than we were before?

Yes, but we need this information to properly do type lowering for closure arguments, so I do not see a way around this without a major redesign. What are your concerns?

Just that IIRC serialization was the major thing blocking us from putting Clang types into the AST in the first place, so if we're serializing them in more places, that gives me pause. I wouldn't have expected this technical approach to require that.

[Xazax-hun]

@swift-ci please test

[kavon]

LGTM