[BFT-A][R9] Implement canonical BUY_CBE issuance per mint-on-buy spec

[umwelt]

Title

[BFT-A][R9] Implement canonical BUY_CBE issuance per mint-on-buy spec

Category

• R9 Negative Spec Enforcement

Authoritative Inputs

• CBE Bonding Curve — Complete Formula Reference section 5
• CBE Bonding Curve — Rust Implementation Specification section 5
• Team decision: GRAD_THRESHOLD = 2_745_966 * SCALE

Problem

The current issue incorrectly framed SELL_CBE as something to remove. The new docs clearly define BUY_CBE first, with exact reserve/treasury split, buyer/SOV token split, state checkpointing, and post-commit graduation.

Current Behavior

There is no working canonical BUY_CBE path that matches the documented fixed-width transaction, fee model, split rules, slippage behavior, and state transition ordering.

Required Behavior

Implement BUY_CBE exactly as specified:

• fee deduction from buyer balance_sov into dao_treasury_sov
• reserve/treasury split using RHO
• multi-band mint computation using reserve credit only
• expected_s_c stale-state check
• buyer/SOV token split using ALPHA
• compute buyer_tokens via ratio and sov_tokens by subtraction to preserve exact sum
• buyer slippage check using the documented formula:
  • buyer_effective = amount_in * SCALE / buyer_tokens
  • reject if buyer_effective > max_price
• atomic state commit
• graduation check on reserve_balance >= 2_745_966 * SCALE

Safety Impact

Without the documented BUY_CBE path, supply, reserve, treasury, and graduation semantics remain undefined.

Proposed Fix

Implement BUY_CBE in the dedicated curve executor lane and remove old shortcut/approximation behavior.

Dependencies

• Depends on [BFT-A][R9] Replace curve arithmetic with 18-decimal U256 fixed-point math and deterministic integer sqrt #1924
• Depends on [BFT-A][R8] Collapse CBE curve math to one canonical piecewise-linear engine #1925
• Depends on [BFT-A][R8] Add dedicated CBE curve executor lane and remove disabled legacy dispatch #1929

Acceptance Criteria

• BUY_CBE works end to end from fixed-width bytes to atomic state commit
• Reserve and treasury credits sum exactly to amount_in
• Buyer and SOV token splits sum exactly to total_delta_s
• Fee deduction and DAO treasury credit match the documented sequence
• expected_s_c, max-price slippage, supply limit, and zero-mint checks are enforced
• Graduation is triggered when reserve_balance >= 2_745_966 * SCALE

Test Plan

• Zero-input rejection
• Stale-state rejection
• Slippage rejection using the documented effective-price formula
• Exact mint vectors
• Max-supply rejection
• Graduation threshold transition tests using 2_745_966 * SCALE

Cleanup After

• Remove old buy-path semantics that bypass reserve split, checkpointing, buyer/SOV token distribution, or the documented slippage formula

[github-actions]

The GitHub issue proposes implementing a canonical BUY_CBE issuance path for a bonding curve mechanism while removing the existing native sell/redemption functionality. Currently, the bonding curve lacks a functional buy path (disabled) and still retains a sell path structurally, which leads to undefined supply and reserve semantics. To address this, the proposed solution entails implementing the BUY_CBE operation with atomic updates to state parameters like supply, reserve, and treasury, while enforcing checks for max supply and slippage. The sell path would be hard-rejected across the codebase, with related semantics, documentation, and testing removed. Dependencies for this fix include three other issues (#1924, #1925, #1929), and thorough testing is required to ensure proper rejection behaviors and enforcement of conditions. The cleanup phase would demote the sell mechanism to non-native status and eliminate all associated references from implementation.