Add comprehensive test coverage for boost/stake system

[prodbycorne]

Overview

The boost/stake system (stake, unstake, set_boost, set_global_multiplier, get_credits) is undertested. The core invariant — that credits earned under the old boost config are preserved when the boost changes — is not verified.

Missing Test Cases

compute_total_stake (unit test the private fn via public interface)

• No boost (allocation_pct = 0): total_stake == amount
• 50% allocation, multiplier 2: total_stake == amount * 1.5
• 100% allocation, multiplier 3: total_stake == amount * 3
• 100% allocation, multiplier 1: total_stake == amount (no gain)

set_boost

• allocation_pct = 0 rejected
• allocation_pct = 101 rejected
• Changing boost checkpoints credits earned at old rate (no credit loss)
• boost_applied event emitted with (user, allocation_pct, multiplier)

set_global_multiplier

• Non-admin rejected
• multiplier = 0 rejected
• Changing multiplier does not retroactively alter already-banked credits
• mult_set event emitted

get_credits

• Returns 0 for unstaked user
• Increases monotonically with ledger advancement
• Boost increases credits relative to unboosted at same amount

Credit preservation across boost changes

• Stake 1000 → advance 500 ledgers → set_boost 50% → advance 500 ledgers
  → credits = (unboosted * 500) + (boosted * 500)
  (NOT retroactively recalculated for first 500 ledgers)

Acceptance Criteria

• All cases above have passing tests
• compute_credits arithmetic verified with exact expected values
• get_credits vs calculate_credits distinction documented in test comments
• Full stake/unstake round-trip tested with token balance assertions

[JuanVP2]

Hi @prodbycorne,

I would love to take on this issue to expand the test coverage for the boost and stake systems.

I have already analyzed the required test cases and local invariants. I am ready to implement the complete suite covering:

• Arithmetic unit tests for compute_total_stake boundary conditions (0%, 50%, 100% allocations).
• Validation and error-handling checks for set_boost (0 and 101 bounds) and set_global_multiplier (zero check).
• State-preserving integration tests verifying credit checkpoints across active boost configuration changes (ensuring calculations are never retroactively overwritten).
• Comprehensive token-balance validation using full stake/unstake round-trips.

Since I already have my local Soroban and Rust environment fully configured and passing the current tests, I can deliver a clean, well-documented implementation promptly.