Clarification may be useful in flow-analysis.md

[eernstg]

The flow-analysis definition of toi_promote has the following:

• toi_promote(declared, promotionChain, tested, written), where ...
  ...
  • If the written type is in p2, and written <: provisionalType, then
    newPromotionChain is [...promotionChain, written]. ...
  • Otherwise (when written is not in p2):

This is somewhat confusing to read because it seems to skip over the case where written is in p2, but written <: provisionalType does not hold.

However, written <: provisionalType is guaranteed to hold in all cases because of the preconditions which are specified in order to use toi_promote (and because the only "call site" for toi_promote is known to satisfy those preconditions).

Granted, this is a request for a tiny change, but I do think the rule will be easier to read if "and written <: provisionalType" is removed, or it is changed to commentary.

[eernstg]

A proposal is here: #4507. It simply removes the condition, because it doesn't seem to be necessary to even mention it in commentary, the existing commentary already contains the required information.