Implement more descriptive pretty stack printing for generated derivatives.

[vgvassilev]

This prints:

1.	<eof> parser at end of file
2.	Building code for '<double wrapper_function(double x, double (*func)(double))>[name=wrapper_function, order=1, mode=pushforward, args='', tbr]'
 #0 0x0000000104d4c680 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/Users/vvassilev/workspace/builds/llvm-project-clad/bin/clang-18+0x102c14680)
 #1 0x0000000104d4cc64 PrintStackTraceSignalHandler(void*) (/Users/vvassilev/workspace/builds/llvm-project-clad/bin/clang-18+0x102c14c64)
 #2 0x0000000104d4a91c llvm::sys::RunSignalHandlers() (/Users/vvassilev/workspace/builds/llvm-project-clad/bin/clang-18+0x102c1291c)
 #3 0x0000000104d4dd80 SignalHandler(int) (/Users/vvassilev/workspace/builds/llvm-project-clad/bin/clang-18+0x102c15d80)
 #4 0x0000000185424624 (/usr/lib/system/libsystem_platform.dylib+0x1804ac624)
...

@PetroZarytskyi, can you help me put what would help us make more sense out of a stack trace? I thought it might be a good idea to print also the last statement of the forward and reverse passes. Would that help?

[github-actions]

clang-tidy made some suggestions

[codecov]

Codecov Report

❌ Patch coverage is 34.28571% with 23 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
lib/Differentiator/VisitorBase.cpp	0.00%	21 Missing ⚠️
...clude/clad/Differentiator/BaseForwardModeVisitor.h	80.00%	1 Missing ⚠️
include/clad/Differentiator/ReverseModeVisitor.h	66.66%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[PetroZarytskyi]

Probably not for this PR, but I think we need to include DVI in the DiffRequest dump. I did it locally while working on static graph PRs as it's crucial for matching requests and caused the most trouble.

[PetroZarytskyi]

This functionality looks very useful. At first, I thought it would be amazing to see the last original statement we started visiting (ideally, with location). This would be enough info for 80% of the cases when I need to run debuggers. But perhaps this would be hard to implement. Dumping the last generated stmts seems like a great idea as it does pretty much the same job.

[vgvassilev]

This functionality looks very useful. At first, I thought it would be amazing to see the last original statement we started visiting (ideally, with location). This would be enough info for 80% of the cases when I need to run debuggers. But perhaps this would be hard to implement. Dumping the last generated stmts seems like a great idea as it does pretty much the same job.

Did you mean:

1.	<eof> parser at end of file
2.	Building code for '<double wrapper_function(double x, double (*func)(double))>[name=wrapper_function, order=1, mode=pushforward, args='', tbr]'
While visiting <IntegerLiteral> [ '/Users/vvassilev/workspace/sources/clad/test/Misc/Test.cpp:11:22', '/Users/vvassilev/workspace/sources/clad/test/Misc/Test.cpp:11:22']

Begin Stmt Dump ---
2
End Stmt Dump ---

[github-actions]

clang-tidy made some suggestions

[github-actions]

warning: member variable 'm_CurVisitedStmt' has protected visibility [cppcoreguidelines-non-private-member-variables-in-classes]

    const clang::Stmt* m_CurVisitedStmt = nullptr;
                       ^

[vgvassilev]

This is tested but only run in debug mode. Our codecov infrastructure does not do run in debug mode that's why it complains.