Fix missing PaymentSent::fee_paid_msat

[valentinewallace]

If an outbound payment was abandoned with htlcs in-flight and later claimed, we would previously have the PaymentSent::fee_paid_msat be set to None. This contradicted some docs on the event that stated the field would always be Some after 0.0.103.

Fixes #4639

[ldk-reviews-bot]

👋 Thanks for assigning @TheBlueMatt as a reviewer!
I'll wait for their review and will help manage the review process.
Once they submit their review, I'll check if a second reviewer would be helpful.

[ldk-claude-review-bot]

Review Summary

Issue found

• lightning/src/ln/outbound_payment.rs:169 — remove() does not decrement pending_fee_msat for the Abandoned variant, causing fee_paid_msat to be overstated for MPP payments where some paths fail after manual abandonment.

Details

The core fix (preserving pending_fee_msat through the Retryable → Abandoned transition) is correct and works for single-path payments and MPP payments where all paths succeed. However, the remove() method (line 335) only adjusts pending_fee_msat for Retryable, not Abandoned. When a path fails on an already-abandoned MPP payment, the fee for that failed path is never subtracted, leading to an incorrect (overstated) fee_paid_msat in the PaymentSent event.

The new test only covers a single-path payment, so it doesn't exercise this edge case.

[valentinewallace]

Fuzz failure looks pre-existing, seems to error with "Expected disconnect"

[ldk-reviews-bot]

🔔 1st Reminder

Hey @tankyleo! This PR has been waiting for your review.
Please take a look when you have a chance. If you're unable to review, please let us know so we can find another reviewer.

[tankyleo]

In this block should we also subtract path.final_value_msat() from total_msat so that PaymentSent reports the amount that was actually sent ?

[valentinewallace]

I removed this block. If we go in this direction, we're starting to handle surfacing correct amounts/fees in the case that the recipient gave us the preimage without us paying them the full amount we originally intended. IIUC this indicates a very buggy recipient and is not a case we handle in PendingOutboundPayment::Fulfilled either, so a more "real" fix would be a larger change and probably not worth it. Similar applies to the below comment.

[tankyleo]

curious here if we should be setting this to the sum of the current paths currently in-flight, and not the fixed total msat ?

[ldk-claude-review-bot]

Bug: remove() (line 335) only decrements pending_fee_msat for the Retryable variant, not for Abandoned. This means if an MPP payment is manually abandoned with multiple paths in-flight, and then some paths fail back while others succeed, the fee for the failed paths won't be subtracted — causing PaymentSent::fee_paid_msat to be overstated.

Concrete scenario:

1. Two-path MPP, path A (fee=100) and path B (fee=200) in-flight → pending_fee_msat = Some(300)
2. abandon_payment() → Abandoned { pending_fee_msat: Some(300) }
3. Path A fails (intermediate node timeout) → remove() removes session_priv but doesn't decrement fee → pending_fee_msat still Some(300)
4. Path B succeeds (recipient claims) → PaymentSent { fee_paid_msat: Some(300) } — should be Some(200)

The remove() function at line 348-363 needs an additional branch for Abandoned:

if let PendingOutboundPayment::Abandoned {
    ref mut pending_fee_msat, ..
} = self {
    let path = path.expect("Removing a failed payment should always come with a path");
    if let Some(fee_msat) = pending_fee_msat.as_mut() {
        *fee_msat -= path.fee_msat();
    }
}

Note: for the auto-abandon path (in handle_update_fail_htlc), remove() is called before mark_abandoned(), so the fee is correctly captured. This bug only affects the manual-abandon path and any future failures arriving after auto-abandonment.

[valentinewallace]

See #4651 (comment)