Model system_libs in cpp_info/deps_cpp_info and generators

[danimtb]

Changelog: Feature: Add system_deps attribute for cpp_info and deps_cpp_info.
Docs: conan-io/docs#1395

• Refer to the issue that supports this Pull Request: related to cppinfo improvements: components, exe, system deps... #5090, closes System library dependencies could be specified in the wrong order #2382
• If the issue has missing info, explain the purpose/use case/pain/need that covers this Pull Request.
• I've read the Contributing guide.
• I've followed the PEP8 style guides for Python code.
• I've opened another PR in the Conan docs repo to the develop branch, documenting this one.

^{Note: By default this PR will skip the slower tests and will use a limited set of python versions. Check here how to increase the testing level by writing some tags in the current PR body text.}

[memsharded]

I think this PR is ok, but it needs to be more complete, to actually use that value in the generators. I'd like to see first how this value is going to be used by consumers.

[lasote]

So, this is in the same status @memsharded already commented right?
The cmake generator is creating a variable but not even for cmake the vars are going to be linked.
For other generators the same. These libs need to be used.

[danimtb]

Sorry that I didn't comment after the change. Yes, the changes were made. Previously system_deps was only added to the cpp_info model. Now a variable in the cmake generator is created and ready to be consumed

[lasote]

But ready to be used how... Shouldn't we add them to the targets in right order and into the global CONAN_LIBS or am I missing something?

[lasote]

And same for other generators, In my opinion, we cannot release this if it is not going to work out of the box.

[danimtb]

The system_deps libraries are appended to CONAN_LIBS in the cmake generator so it should work in the same way. This PR also includes a cmake variable called CONAN_SYSTEM_DEPS_{dep}{build_type} to differentiate it from the package libs

[lasote]

Missing tests for:

• CMake target approach
• CMake find_package approach
• CMake find_package_multi approach
• Other generators.

[memsharded]

Please have a look a the comments, specially the one about a changed test and if it could be breaking.

[memsharded]

We need to make this backwards compatible, and also consistent with what is done to the global aggregated variables. If CONAN_LIBS is showing package libs + system libs + frameworks, then CONAN_LIBS_MYPKG needs to show the same for MyPkg.

The way to be able to introduce the changes to the package variables is to define new ones as necessary:

• CONAN_PKG_LIBS_MYPKG: Contains only the package defined ones (MyPkg)
• CONAN_SYSTEM_LIBS_MYPKG: Contain only the system ones (of MyPkg)
• CONAN_FRAMEWORKS_MYPKG: Contain the frameworks (of MyPkg)
• CONAN_LIBS_MYPKG: Contains the addition of all (MyPkg). Final result should remain the same as in the past.

[danimtb]

Changes are done as requested in the review. I have not added the frameworks variables to avoid making more noise