docs: document Karatsuba attempt and why schoolbook is kept for 8 limbs

Karatsuba multiplication (splitting 8 limbs into 4+4 halves for 48 inner
products instead of 64) was implemented and tested but reverted because
the overhead of extra additions, carry propagation, and 5 temporary array
allocations per call negates the product-count savings at only 8 limbs.
The crossover point where Karatsuba beats schoolbook is typically ~32+
limbs on hardware with fast multiply.

https://claude.ai/code/session_01BhU63WUe9AhikZxRdw3Lpg

quartz/src/commonMain/kotlin/com/vitorpamplona/quartz/utils/secp256k1/U256.kt

@@ -139,6 +139,10 @@ internal object U256 {
      * Uses the standard O(n²) algorithm with 8×8 = 64 inner Long multiplications.
      * Each partial product is at most 32×32 = 64 bits, which fits in a signed Long
      * with room for carry accumulation.
+     *
+     * Note: Karatsuba (splitting into 4-limb halves for 48 products) was attempted
+     * but the overhead of extra additions, carry propagation, and 5 temporary array
+     * allocations per call negates the product-count savings at only 8 limbs.
      */
     fun mulWide(
         out: IntArray,