[codex] prove same-block reserve mint removes pending dilution

[mejango]

What changed

Adds regression tests and an implementation fix for same-timestamp reserve minting before scorecard submission.

DefifaHook now records lastReserveMintTimestamp when reserves are minted. DefifaGovernor.submitScorecardFor rejects scorecard submission in the same timestamp as the most recent reserve mint, then allows submission in the next timestamp with the reserve holder included normally.

What this fixes

A same-block reserve mint before scorecard submission could remove pending-reserve dilution from the denominator while the historical checkpoint still excluded the newly minted reserve. That could inflate the paid holder's BWA.

The new ordering guard closes the same-timestamp path without broadly disabling reserve minting during scoring.

Validation

forge test --root defifa --match-path test/regression/FixPendingReserveDilution.t.sol

Result: 4 passed.

forge build --deny notes --skip "*/test/**" --skip "*/script/**"

Result: passed.

forge test --deny notes --fail-fast --summary --detailed --skip "*/script/**"

Result: passed. Full local suite completed with no failures.

[mejango]

This has not been deemed severe to the point where a redeploy is recommended.