support BIP342 script-path taproot sighash

[fmoletta-muun]

hashForTaprootSignature only handled key-path (BIP341) spends. Allow a non-null scriptCode, interpreted as the 32-byte tapleaf hash, to produce the script-path sighash: set spend_type to 0x02 and append the ext_flag=1 extension (tapleaf_hash || key_version || codesep_pos). Key-path behaviour is unchanged.

Add testSigHashTaprootScriptPath, whose golden value is cross-checked against btcd's txscript.CalcTapscriptSignaturehash (the same oracle reproduces the existing key-path vector exactly).

[LeanSerra-muun]

This test is failing because the value of 2_236_761L is incorrect (with 2_236_249L the test passes). Maybe we can reuse the parsing from the test that is above this one:

bitcoinj/core/src/test/java/org/bitcoinj/core/TransactionTest.java

Lines 484 to 504 in 67d8a58

	final List<TransactionOutput> prevOuts = new ArrayList<>();
	new Message(RegTestParams.get(), Hex.decode(prevOutsHex), 0, 0) {
	@Override
	protected void parse() throws ProtocolException {
	this.length = 1;
	final long length = readVarInt();
	for (long i = 0; i < length; i++) {
	final BigInteger value = readUint64();
	final byte[] script = readByteArray();
	prevOuts.add(new TransactionOutput(
	RegTestParams.get(),
	null,
	Coin.valueOf(value.longValue()),
	script
	));
	}
	}
	};

[fmoletta-muun]

Fixed by using the parsing logic 8701562

[LeanSerra-muun]

Considering we don't modify this code often and for future readability I think we should modify the argument scriptCode to tapLeafHash. Also I would add a nice doc comment to this function documenting that we don't support OP_CODESEPARATOR and what each argument is similar to what the library has for other functions:

bitcoinj/core/src/main/java/org/bitcoinj/core/Transaction.java

Lines 1302 to 1317 in 67d8a58

	}
	/**
	* <p>Calculates a signature hash, that is, a hash of a simplified form of the transaction. How exactly the transaction
	* is simplified is specified by the type and anyoneCanPay parameters.</p>
	*
	* <p>This is a low level API and when using the regular {@link Wallet} class you don't have to call this yourself.
	* When working with more complex transaction types and contracts, it can be necessary. When signing a Witness output
	* the scriptCode should be the script encoded into the scriptSig field, for normal transactions, it's the
	* scriptPubKey of the output you're signing for. (See BIP143: https://github.com/bitcoin/bips/blob/master/bip-0143.mediawiki)</p>
	*
	* @param inputIndex input the signature is being calculated for. Tx signatures are always relative to an input.
	* @param scriptCode the script that should be in the given input during signing.
	* @param prevValue the value of the coin being spent
	* @param type Should be SigHash.ALL
	* @param anyoneCanPay should be false.

[fmoletta-muun]

Added doc here 52b1f21