test(python/evm): add unit tests for multicall module

[0xAxiom]

What

Add 32 unit tests for python/x402/x402/mechanisms/evm/multicall.py — the largest EVM mechanism source file in the Python SDK without a dedicated unit test module.

Why

multicall.py is exercised indirectly through eip3009_utils paths but has no module-level test file. This PR fills that gap with focused tests that pin the public API and the key normalization edge cases.

Coverage

MULTICALL3_ADDRESS / MULTICALL3_TRY_AGGREGATE_ABI (3 tests)

• Canonical Multicall3 address
• ABI describes a tryAggregate function
• Inputs match (bool, tuple[](address,bytes))

encode_contract_call (5 tests)

• Selector + ABI-encoded args for balanceOf(address)
• Selector-only output for no-arg function
• Tuple input canonicalization (submit((address,uint256)))
• ValueError when function name not in ABI
• Skips non-function ABI entries (events, etc.)

multicall typed path (5 tests)

• Decodes single uint256 output
• Decodes single string output
• Returns a list for multi-output function
• Returns None for function with no outputs
• ABI decode failure marks result unsuccessful with error populated

multicall raw call_data path (3 tests)

• Skips decoding and signals success=True with result=None
• Raw bytes propagate verbatim into aggregate calls
• (False, b"") maps to RuntimeError("multicall: call reverted")

Validation guards (4 tests)

• Empty calls list short-circuits without calling the signer
• Typed entries missing ABI raise ValueError
• Typed entries missing function_name raise ValueError
• MULTICALL3_ADDRESS/tryAggregate/require_success=False propagation

Result-shape normalization (_normalize_results, 6 tests)

• Dict entries with hex-string returnData
• Objects with success / returnData attributes
• Plain (bool, bytes) tuple entries
• Rejects non-sequence response
• Rejects entry of unexpected shape
• Rejects non-bytes returnData

Mixed ordering & failures (2 tests)

• Mixed typed + raw calls preserve input order
• Revert on typed call returns success=False without attempting to decode

Length mismatch (2 tests)

• Too few results raises ValueError
• Too many results raises ValueError

MulticallResult dataclass (2 tests)

• Default values for failed-with-no-error case
• Error propagation

Test Run

$ uv run pytest tests/unit/mechanisms/evm/test_multicall.py -q
................................                                         [100%]
32 passed, 1 warning in 1.58s

$ uv run pytest tests/unit/ -q
843 passed, 1 warning in 3.48s

Checklist

• No production code changes — pure unit tests
• uvx ruff format clean
• uvx ruff check clean
• All 843 Python unit tests still pass
• Changelog fragment added (changelog.d/multicall-unit-tests.doc.md)
• Commit GPG-signed

[cb-heimdall]

🟡 Heimdall Review Status

Requirement	Status	More Info
Reviews	🟡 0/1	Denominator calculation Show calculation 1 if user is bot 0 1 if user is external 0 2 if repo is sensitive 0 From .codeflow.yml 1 Additional review requirements Show calculation Max 0 0 From CODEOWNERS 0 Global minimum 0 Max 1 1 1 if commit is unverified 0 Sum 1