🐛 [BUG] - chain.connect(fork=None) silently runs without fork

[0xInnominatus]

Description

When @chain.connect(chain_id=NNN,fork=...) is given None or an empty string (e.g. fork=os.getenv("ETHEREUM_RPC_URL") when the env var is unset), Wake still starts the dev chain without a fork. Anvil is launched with only --chain-id (and other flags), but no -f.

That is easy to miss: tests that assume mainnet state (real contracts at known addresses, specified chain_id, etc.) run against an empty chain and fail with confusing errors in the middle of test.

Error is always ValueError: non-hexadecimal number found in fromhex() arg at position 1 and some deploys and txs have passed in test, so it's unclear where probleam appears.

Should Wake fail fast when fork is None or empty? Optionally, could it also warn when chain_id is set but no fork URL was provided - to catch misconfigured fork tests? Also it is possible to make error more clear?

Reproduction steps

1. Clone reproduction repo with MRE
2. Compile contracts with eth-wake (`wake compile`)
3. `wake up pytypes`
4. `wake test`

Reproduction URL

https://github.com/0xInnominatus/eth-wake-mre

Screenshots

![DESCRIPTION](LINK.png)

Logs

def test_default():
        token = ERC20.deploy("Example", "EXAMPLE", 100 * 10**18)
>       example = Example.deploy(WETH)

tests\test_default.py:18:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
pytypes\contracts\Example.py:89: in deploy
    return cls._deploy(request_type, [token], return_tx, Example, from_, value, gas_limit, {}, chain, gas_price, max_fee_per_gas, max_priority_fee_per_gas, access_list, type, block, confirmations)
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\core.py:3246: in _deploy
    return cls._execute(
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\core.py:3381: in _execute
    return chain._transact(
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\core.py:1429: in wrapper
    return f(*args, **kwargs)
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\core.py:2820: in _transact
    if tx.error is not None:
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\transactions.py:85: in wrapper
    return f(self)
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\transactions.py:450: in error
    raw_error = self.raw_error
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\transactions.py:85: in wrapper
    return f(self)
..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\transactions.py:498: in raw_error
    revert_data = bytes(read_from_memory(offset, length, trace["memory"]))
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

offset = 0, length = 0
memory = ['0x0000000000000000000000000000000000000000000000000000000000000000', '0x00000000000000000000000000000000000000000000...0000000c02aaa39b223fe8d0a0e5c4f27ead9083c756cc2', '0x06fdde0300000000000000000000000000000000000000000000000000000000']

    def read_from_memory(offset: int, length: int, memory: List) -> bytearray:
        start_block = offset // 32
        start_offset = offset % 32
        end_block = (offset + length) // 32
        end_offset = (offset + length) % 32

        if start_block == end_block:
            if start_block >= len(memory):
                return bytearray(length)
>           return bytearray.fromhex(memory[start_block])[start_offset:end_offset]
E           ValueError: non-hexadecimal number found in fromhex() arg at position 1

..\..\..\AppData\Local\Programs\Python\Python312\Lib\site-packages\wake\development\internal.py:26: ValueError

OS

Windows, Linux

[husanGuru]

I hit the exact same traceback (wake/development/internal.py:26, read_from_memory(), ValueError: non-hexadecimal number found in fromhex() arg at position 1) but I don't think it's actually caused by the fork=None issue described here. I can reproduce it on a plain, non-forked local Anvil chain with no fork configuration involved at all.

read_from_memory() only parses the memory array from debug_traceTransaction's structLogs, so it shouldn't matter whether the chain is forked or not. In my case the actual cause is that Anvil >=1.7.0 now returns each 32-byte memory word 0x-prefixed (e.g. "0x0000...0080" instead of "0000...0080"), which breaks the unconditional bytearray.fromhex(memory[i]) call.