Introduce support for CMake-based ns-3 versions and fix tests

[pagmatt]

This PR completes the support for the newer, CMake-based versions of ns-3.

The main changes are:

• Updated CI workflow checkout action from v2 to v3 to solve a related warning.
• Switched to ns-3 mainline repository for the examples/ns-3 submodule. In order to test SEM with both the CMake and the waf versions of ns-3, in this case the repo is checked out at release ns-3.35
• When merging TinyDB tables, re-assign document IDs (which must be unique for each entry) in order to prevent errors. In this regard, the additional import of Document from tinydb.table has been added in cli.py
• Introduced support for platform-dependent build status filename in ns-3.36+
• In the configure_and_build function, do not redirect to PIPE stderr when calling subprocess.Popen ().
  When compiling, you can now get a mix of stdout and stderr output when compilation errors and/or warnings arise. This was causing a race condition between stdout.readline () and subprocess.poll () which caused get_build_output to never return.
  It is actually suggested to use subprocess.communicate () as a safer alternative; however, this would not allow us to retrieve the compilation progress. Thus, the proposed solution of missing out on stderr, while at least capturing stdout.
• Prevented copying CMake cache when creating symlinks in the tests. Previously, this caused errors since the cache is path-dependent. I tried to circumvent this dependence instead of not copying the cache, but I was not able to do it.
• Enabled only the core module by using the --enable-modules=core flag when configuring ns-3 during the tests. Since only such module is needed, this was an obvious choice as it significantly reduces testing time.
• In the tests, switched from /usr/share/dict/web2 and /usr/share/dict/web2a as input parameters to /usr/share/dict/american-english and /usr/share/dict/british-english. The previous dictionaries are not installed by default on recent Ubuntu distributions.

[MatteoDrago]

Massive work, @pagmatt.

Shouldn't get_command_from_result from sem/utils.py be updated for consistency?

[pagmatt]

Massive work, @pagmatt.

Shouldn't get_command_from_result from sem/utils.py be updated for consistency?

Thanks @MatteoDrago!
Indeed it should.
My plan is to first release an updated version of SEM which at the very least works, and then work on fixing smaller issues, such as the one you pointed out! But I will try to sneak this one in 😄

[MatteoDrago]

Massive work, @pagmatt.
Shouldn't get_command_from_result from sem/utils.py be updated for consistency?

Thanks @MatteoDrago! Indeed it should. My plan is to first release an updated version of SEM which at the very least works, and then work on fixing smaller issues, such as the one you pointed out! But I will try to sneak this one in 😄

Makes totally sense. As a matter of fact, the doc should be updated as well in some points... But that's for another PR, I agree! Thanks!