Report coverage for tests

[LucaGuerra]

Motivation

In the last 1~2 years we have improved test suites and coverage by a lot. If you look at libsinsp unit testing alone we went from about 6% to ~70% of the library. In addition we have the Python e2e suite and the recent C++ e2e suite, plus all driver tests that exercise single syscalls. The amount of coverage we get from all those should be maintained going forward so we should measure it and report it on PRs for each large test suite.

How to properly measure that in a way that is not misleading may be tricky and require experimentation

Alternatives

Leave it unreported. People will still add tests but maybe sometimes we won't have full context.

Additional context