fix(docs): correct UINT_40_MAX bit width (48 -> 40) in README

[MitkoTschimev]

Summary

The inline example in the README (line 28) defines UINT_40_MAX with the wrong bit width:

// Current (wrong — this is UINT_48_MAX)
const UINT_40_MAX = (1n << 48n) - 1n

// Fixed (correct 40-bit max)
const UINT_40_MAX = (1n << 40n) - 1n

Evidence

• @1inch/byte-utils defines UINT_40_MAX = (1n << 40n) - 1n and UINT_48_MAX = (1n << 48n) - 1n as separate constants
• MakerTraits.withNonce() docs state: "nonce must be less or equal to uint40::max"
• The RFQ Order example on line 65 of the same README correctly imports UINT_40_MAX from @1inch/byte-utils

Impact

The wrong value happens to work in practice because the SDK's setMask operation truncates, but it is semantically incorrect and could mislead developers or cause issues if stricter validation is added.