Second pass with the avoid-ai-writing detector

The skill's detector scored the README, PARITY.md and the extracted
comment prose. Fixed what it flagged: hollow intensifiers, "in order
to", "additionally", "the whole point", bold sentences in docs, and
bold that had wrapped across comment lines. Comment prose scores a
maximum of 4 out of 100 across the repo.

One sweep rule mangled four comments ("which is the entire point of X"
lost its clause); repaired, and the rule now keeps the clause.

Comments and strings only. 1208 tests pass, counted from the XML.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

engine/src/test/kotlin/io/flamenet/e2e/HdAccountTest.kt

@@ -10,8 +10,8 @@ import kotlin.test.assertTrue
 /**
  * From a recovery phrase to an address somebody can be paid at.
  *
- * **The property that matters most here is that the signing key and the
- * displayed address come from the same derivation.** If they ever disagreed,
+ * The property that matters most here is that the signing key and the
+ * displayed address come from the same derivation. If they ever disagreed,
  * the wallet would sign with a key that does not own the address it showed. The
  * transaction being rejected is the GOOD outcome; the bad one is signing for an
  * address somebody else controls.
@@ -160,8 +160,8 @@ class HdAccountTest {
     }
 
     /**
-     * **The limit, asserted as a passing test so nobody reads
-     * `looksLikeAddress` as stronger than it is.**
+     * The limit, asserted as a passing test so nobody reads
+     * `looksLikeAddress` as stronger than it is.
      *
      * A Solana address carries NO checksum. A mistyped character that still
      * decodes to 32 bytes is a perfectly valid address that nobody holds the