Adapting Ravenbrook and MM group review procedure to public MPS

[rptb1]

Introduces Memory Pool System review procedure in order to solve #101 .

Work stages:

1. Review all the documents mentioned in MPS documentation lacks a review procedure #101 (comment) and copy the relevant parts of their text to the procedure, in order to capture all the process improvement value in those.
2. Update the documents to use more current language ("pull request", "issue comment", etc.) and current tools (Git, GitHub).
3. Test the procedure.
4. Subject the procedure to itself.

[rptb1]

The second test run of procedure at #83 (comment) went smoothly. No real problems found in the procedure, but a few good suggestions.

Note for inclusion: Base kickoff, check,log, and brainstorm time seems to be about 2h elapsed for the team we have. The procedure itself suggests a limit of 3.5h. Express review is capped at 30m.

[rptb1]

Consider renaming "review" to "inspection" for this particular process:

• it's more accurate
• it doesn't co-opt the word "review" which is useful for lots of other things
• it's less likely to confuse people used to other kinds of review
• it makes the point to outsiders that we're doing something more than the usual review
• it might break references from older material a bit, including the old MM stuff.

[rptb1]

Some thoughts about how to apply this document to itself, when it comes to it. It's about 3600 lines of text, so that's 6h at the baseline 10 lines/min. However, it's quite reviewable in sections, especially since most of the procedures a made to be done separately.

Consistency with the introduction and rationale is important. And the overall structure of the document for its purpose will need attention, possibly both at the start (to see how it appeals to someone new approaching it) and at the end (when everyone's read it all).

We can also switch tactics after one or two sessions.

[rptb1]

In #119 (review) @thejayps mentioned design/guide.review.txt, guide.impl.c.format, and guide.impl.c.naming . These are Ravenbrook-era review documents that need absorbing/merging into rules and checklists.

[thejayps]

Who is responsible for assigning non checking roles (eg scribe, pi) and when does this happen?

Currently it is not happening on some reviews. This is related to #274

[thejayps]

After looking at "something random" in weekly meeting: https://info.ravenbrook.com/rule/ was found as a source document for review of this