[ANN] netdev stats and patch scoring now credit the patch author

[Jakub Kicinski <kuba@kernel.org>]

Hi!

We're experimenting with/changing how we count patch authorship
in ml-stats. tl;dr is that we used to always count patches as
authored by the poster, now we extract authorship from within
the email, like git does.

Before
------

Previously everything in a posting counted for the sender, as an author.
Everyone else replying in the thread counted as a reviewer. 
Pull requests did not count as postings at all.

The motivation was to place the person acting as a maintainer
as a representative. For NIC vendors especially, I thought
it would be unrealistic to get all the authors to review
code upstream. Conversely the maintainers should follow the list
and bring the knowledge back to internal review.

Sadly, none of the vendor "representatives" acted on those incentives.
Perhaps it's Planck's principle at work, who knows.

The old system while trying to "incentivise" vendor representative
participation was unfairly punishing sub-maintainers (netfilter,
can, mptcp etc.) Instead of trying to separate the two groups
I decided to change the accounting.

After
-----

The patches now count for their author. Whoever sent them is still 
on the "submitting side" of the thread - their replies there are not
reviews - but they are only charged for what they wrote themselves,
like the cover letter. Authors whose patch someone else posted are
authors of that thread, not reviewers.

The "author" ranking shuffles quite a bit as all sub-maintainers 
drop out. But the top 15 individual score table does not change.
Pengutronix drops out of the negative score table since all CAN
submissions are no longer counted as Marc's work.

This change will obviously impact SUIE scores and therefore processing
order. We will keep an eye on the scores and adjust the logic as necessary.
I'm not entirely sure what to expect there, TBH.