fix: replace speculative resolve() with mark_voted() to prevent dupli…

[enjoyandlove]

Summary

Closes #343

• tracker.resolve() was being called speculatively in the forward loop after a confirm or timeout vote succeeded. resolve() removes the swap from voted_ids, destroying the duplicate-vote guard until event_watcher fires.
• Both call sites (line 408 confirm_miner_fulfillments, line 520 enforce_swap_timeouts) are replaced with tracker.mark_voted(), which records the vote without dropping the guard.
• event_watcher continues to call tracker.resolve() authoritatively on SwapCompleted/SwapTimedOut chain events — that ownership boundary is now respected.

Why

Between "our vote lands" and "event_watcher observes the finality event" (~12 s per forward step, typically several steps of lag), is_voted() was returning False and a redundant extrinsic was submitted on every step. This burned TAO fees on every affected swap and flooded logs with "already voted" rejections, masking legitimate failures.

Test plan

• Existing tests/test_swap_tracker.py mark_voted / is_voted / prune_stale_voted_ids round-trip tests pass
• uv run pytest tests/
• Verify that after a successful confirm_swap or timeout_swap the swap is not re-submitted on the next forward step before event_watcher fires

[enjoyandlove]

Hi, @anderdc
I have just submitted this PR.When you have a moment, could you please review it?
Thank you for your time.