Fix precision loss in treasury amount calculation during fee collection

[coderabbitai]

Description

There is a potential precision loss issue in the treasury amount calculation within the MarkFeesCollected method in internal/service/fees.go.

Problem

The conversion from big.Float to uint64 using the Uint64() method truncates the value and ignores the accuracy flag, which could lead to incorrect treasury amounts being calculated.

Location: internal/service/fees.go around lines 162-165

Current problematic code:

treasuryAmount := big.NewFloat(0).SetUint64(fee.Amount)
treasuryAmount = treasuryAmount.Mul(treasuryAmount, big.NewFloat(treasurySplit))
treasuryAmountUint64, _ := treasuryAmount.Uint64()  // <- Ignores accuracy flag
developerAmountUint64 := fee.Amount - treasuryAmountUint64

Solution

The code should check the accuracy returned by Uint64() and apply proper rounding before converting to uint64 to ensure the treasury amount is correctly calculated without truncation errors.

Backlinks

• Related PR: Treasury records #287
• Original comment: Treasury records #287 (comment)

Reported by: @garry-sharp