test: add update_rewards epoch length flow

[arad-starkware]

This change is 

---

Note

Introduce per-block consensus rewards driven by average block duration, add RewardSupplier EIC and config, and wire staking to cache/use epoch block rewards with extensive tests and spec updates.

• Consensus rewards (block-duration based):
  • Add SECONDS_IN_YEAR constant and compute block rewards using average block duration.
• Reward Supplier:
  • New update_current_epoch_block_rewards to compute per-block (STRK/BTC) rewards.
  • Track avg_block_duration, block_snapshot, and BlockDurationConfig with min/max clamping.
  • Add get_block_duration_config and set_block_duration_config (app governor-only).
  • Add new errors for block/avg duration and overflows.
  • Introduce RewardSupplierEIC to initialize new storage (avg/min/max) during V2→V3 upgrade.
• Staking:
  • Cache epoch block_rewards and last_calculated_epoch; compute via RewardSupplier once per epoch.
  • On set_consensus_rewards_first_epoch, pre-initialize next-epoch block rewards.
• Docs/Spec:
  • Document new Reward Supplier APIs, config struct BlockDurationConfig, and error INVALID_MIN_MAX_BLOCK_DURATION.
• Tests/Flows/Utils:
  • Extensive new tests for block-reward path, EIC, config validation, epoch-length effects, and migration flows.
  • Rename/test helpers to use AVG_BLOCK_DURATION; add advance_blocks and custom epoch-time helpers.

^{Written by Cursor Bugbot for commit 4dcf4e2. This will update automatically on new commits. Configure here.}

[arad-starkware]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• test: add update_rewards old delegator flow #94
• test: add update_rewards transition with balance change flow #93
• test: add update_rewards several blocks delegation rewards test #92
• test: add disable_rewards intermittently flow #91
• test: add update_rewards epoch length flow #90 👈 (View in Graphite)
• test: add update_rewards same epoch flow #89
• test: add update_rewards staker intent flow #88 : 1 other dependent PR (#95 )
• test: add update_rewards empty pools flow #87
• test: add update_rewards btc pools flow #86
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cursor]

Bug: Missing epoch advancement before baseline rewards measurement

The test claims default_length_rewards immediately after start_consensus_rewards() without advancing through any epochs where rewards are active. The start_consensus_rewards() function sets rewards to start at current_epoch + K and advances exactly K epochs, arriving at the start of the first rewards epoch with zero accumulated rewards. Therefore default_length_rewards would be 0 or nearly 0. When the test later claims increased_length_rewards after advancing an epoch, this value would be positive. The assertion increased_length_rewards < default_length_rewards would evaluate to positive < 0, which is false and would cause the test to fail.

 

[noa-starkware]

Reviewable status: 0 of 2 files reviewed, 2 unresolved discussions (waiting on @arad-starkware)

---

src/flow_test/test.cairo line 3830 at r2 (raw file):

    let decreased_length_rewards = system.staker_claim_rewards(:staker);

    assert!(increased_length_rewards < default_length_rewards);

assert all not zero

[cursor]

Bug: Test assertions contradict V3 block rewards calculation logic

The test update_rewards_epoch_length_flow_test asserts that changing epoch_length changes per-block rewards (increased_length_rewards < default_length_rewards < decreased_length_rewards). However, in V3 consensus rewards, block rewards are calculated as yearly_mint * avg_block_duration / (BLOCK_DURATION_SCALE * SECONDS_IN_YEAR), where avg_block_duration = time_delta / num_blocks. Since advance_epoch() uses constant AVG_BLOCK_DURATION per block, the ratio time_delta/num_blocks remains constant regardless of epoch_length. The block rewards should be identical for all epoch lengths, making these assertions incorrect for V3 behavior.