This is a follow-up from  which underlined that the code lacks even basic coverage as it allow
us to introduce a costly regression which could have being detected by pylint no-member test.
- enables pylint with most checks temporary disabled, so we have time
to address them gradually
- fixes few minor issues reported by the tool
- adds skips for some known "no-member" errors but avoids adding it to the exclude list, as this
in order to prevent further regressions.
- once landed we can easily address the temporary disabled errors, one by one. In fact this is could
prove as a very good learning experience for newer team members. They can start by removing on
random exclusion and fixing it. As a hint: leave the missing docstrings for the end, some problems
are more important to fix first.
(cherry picked from commit 13ac87ba36)