Port AutoStackingNumber behavior to Coderive and load via CodP-TAC artifacts

[Copilot]

This change addresses the request to move Java-written Coderive class behavior into Coderive source, then execute it through the Java runtime using compiled CodP-TAC output.
It ports a representative AutoStackingNumber API surface and wires usage through .codb loading rather than relying only on Java class implementation.

• Problem alignment

  • Introduces a Coderive-native implementation for behavior previously exercised from Java-side AutoStackingNumber logic.
  • Ensures the Coderive implementation is compiled to CodP-TAC and consumable by Java runtime import resolution.

• Coderive class port

  • Added src/main/cod/std/math/AutoStackingNumber.cod.
  • Implemented core operations used in parity coverage: constructors/helpers (zero, one, minusOne, fromLong, fromDouble), arithmetic (add, subtract, multiply, divide, remainder, negate, abs), scale-by-power-of-two helpers (shiftLeft, shiftRight), compare, and pow.
  • Clarified shift helper semantics as arithmetic scaling (not bit-level shifts).

• Parity harness in Coderive

  • Added src/main/cod/demo/src/main/test/autostackingnumber/AutoStackingNumberPortParity.cod.
  • Encodes expected results for integer and floating-point cases and reports PASS/FAIL per operation.
  • Uses unit-qualified calls (math.AutoStackingNumber.*) for consistent cross-unit resolution.

• Compiled artifact loading path

  • Added generated artifacts/index entries so runtime can load the Coderive class from CodP-TAC output:
    • src/main/cod/demo/src/bin/math/AutoStackingNumber.codb
    • src/main/cod/demo/src/bin/test.autostackingnumber/AutoStackingNumberPortParity.codb
    • src/main/cod/demo/src/idx/test.autostackingnumber.toml

use {math}

share Demo {
    share main() {
        out(math.AutoStackingNumber.add(123456789, 987654321))
        out(math.AutoStackingNumber.pow(2, -2))
    }
}