Define a cred rating methodology

[teamdandelion]

The 2019 OKRs (#7) call for measuring attribution quality via a subjective "cred rating" from contributors.

Coming up with metrics for attribution quality is challenging, since there
isn't a readily available ground truth to compare against. Right now, the best
metric we have is asking the contributors to a project to rate the quality of
our attribution. The "cred rating" in the table below is the average rating
that contributors to a project give its cred attribution, measured on a
10-point scale.

Metric	Goal	Q4	Q1	Q2	Q3	Q4
SourceCred's cred rating	8/10	-
IPFS's cred rating	7/10	-

@sternhenri points out that we should explicitly document the cred rating methodology. This issue tracks that work.