CLI: detect stale pending_swap.json when miner is re-reserved

[LandynDev]

Summary

alw swap now and alw view reservation were treating any local pending_swap.json as live whenever the saved miner had an on-chain reserved_until > now. That conflates "this miner is reserved by anyone" with "our reservation is still live" — so when a swap completed and another user (or another process) re-reserved the same miner, the stale local row was reported as ACTIVE.

Reported case: user completed swap #3, came back later, alw swap now warned about a pending reservation, alw view reservation showed it ACTIVE with ~16 blocks remaining and the old BTC tx attached, but alw view swap 3 --watch correctly said the swap was already resolved.

Approach

New probe_pending_reservation helper in swap_commands/helpers.py decides among five outcomes:

1. our_swap — get_miner_active_swaps has a swap on this miner whose (user_from_address, receive_address) match ours. Strongest signal; survives vote_extend_reservation and into FULFILLED.
2. expired — no swap match and no reservation row.
3. replaced — reservation amounts differ from saved.
4. replaced — amounts match but reserved_until is more than reservation_ttl past our saved value (a single extension can't push it that far, and chained extensions would have shown up as a swap in step 1).
5. ours_active — within tolerance.

Three call sites consume it: swap.py (start-new prompt), view.py (full table), status.py (dashboard row). Each clears pending_swap.json and prints a one-line resolution when the saved state is no longer ours.

Tradeoffs

The amount-tuple match in step 4 is almost unique-per-reservation but technically could collide if a different user re-reserves the same miner with the same exact amounts. The contract doesn't expose Reservation.from_addr, so the amount tuple plus the ttl-bounded drift check is the best disambiguator without a contract change. In practice, step 1 catches every "stale local file" case once a tx is broadcast (the reservation transitions into a swap and the swap match wins). A contract getter for from_addr would close the residual hole — deferring until the collision actually surfaces.

Test plan

• pytest tests/test_probe_pending_reservation.py — 9 cases covering all 5 branches, including the exact stale-after-completed-swap case
• Full unit suite (pytest tests/) green — 330 passed
• ruff format . + ruff check . clean
• Manual reproduction: complete a swap on dev environment, leave pending_swap.json behind, re-reserve same miner with another wallet, verify alw swap now / alw view reservation / alw status all auto-clear instead of showing ACTIVE