Re-measure the branch conventions on 2.1.0 (#915)

Section 6 said the table was measured on a 2.0.0 build, which was true and is now also
incomplete: 2.1.0 changes logarithm behaviour (#902 guards the exponent pull) and abs (#881), so
a reader is entitled to ask whether the branch choices moved with them. All nine values are
unchanged -- sqrt(-4), (-1)^(1/2), ln(-1), i^i, and the five negative-base powers including the
(-8)^(1/3) against (-8)^0.3333333333 trap.

Re-measuring these whenever a release touches a branch is the point of the section, so the
sentence now says that rather than naming one build.

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: Rafael-SOWNet

Sources/AngouriMath/Docs/Contributing/SimplificationContract.md

@@ -214,8 +214,9 @@ Worked examples, each measured.
 
 ## 6. The conventions this library commits to
 
-Branch cuts disagree between systems, so these are measured on a 2.0.0 build rather than assumed. A
-rewrite that changes any of them changes answers.
+Branch cuts disagree between systems, so these are measured rather than assumed — on a 2.0.0 build
+when this table was written, and re-measured unchanged on 2.1.0, which is worth doing whenever a
+release touches a branch. A rewrite that changes any of them changes answers.
 
 | | value | convention |
 |---|---|---|