fix(wallet): make sign()'s sighash guard context-aware

[muhahahmad68]

Fixes #519.

Description

The pre-flight sighash guard in Wallet::sign() checked every PSBT input against a flat set of allowed sighash values without regard to whether that input was actually Taproot. As a result, TapSighashType::Default (0x00) was accepted unconditionally for every input, including Legacy/Segwitv0 ones, where it has no valid ECDSA meaning.

This is currently masked for the built-in software signer, since sign_input()'s guard (introduced in #476) independently rejects it downstream. However, Wallet::sign() dispatches to any TransactionSigner, and the whole point of this pre-flight guard is that individual signers can trust it has already filtered non-standard sighashes.

Notes to the reviewers

Infer Taproot-ness per input from witness_utxo.script_pubkey, consistent with how complete_transaction() determines Taproot-ness for foreign UTXOs, and apply the correct sighash check for that input's context.

Tests added:

• sign_rejects_ecdsa_input_with_taproot_default_sighash — confirms a non-Taproot input carrying TapSighashType::Default is now rejected
• sign_allows_taproot_input_with_default_sighash — confirms a genuine Taproot spend still signs and finalizes normally

Before submitting

• I followed the contribution guidelines
• This PR breaks the existing API

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.96%. Comparing base (6fc6846) to head (c92965e).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #528      +/-   ##
==========================================
+ Coverage   81.91%   81.96%   +0.04%     
==========================================
  Files          25       25              
  Lines        6535     6542       +7     
  Branches      302      300       -2     
==========================================
+ Hits         5353     5362       +9     
+ Misses       1075     1074       -1     
+ Partials      107      106       -1     

Flag	Coverage Δ
rust	81.96% <100.00%> (+0.04%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ValuedMammal]

One suggestion I have for determining whether an input is a p2tr spend is it may be safer and more reliabe to test witness_utxo.script_pubkey.is_p2tr() rather than the tap_* metadata fields (at least that's consistent with how the check is handled elsewhere e.g. complete_transaction).

[muhahahmad68]

One suggestion I have for determining whether an input is a p2tr spend is it may be safer and more reliabe to test witness_utxo.script_pubkey.is_p2tr() rather than the tap_* metadata fields (at least that's consistent with how the check is handled elsewhere e.g. complete_transaction).

Good catch

[ValuedMammal]

I opened a draft PR to this branch muhahahmad68#1, still open for discussion.
The motivation is to handle the shared logic when it comes to prevout resolution
and input validation. For instance in the allow_all_sighashes guard, we were
determining P2TR-ness by looking at the witness_utxo alone with no additional
validation.