feat(nostr): publish the active ticket wave, not the roll-up

Since upstream v1.6.8 price, currency and inventory belong to time-boxed
ticket waves, and the event-level fields `sync_event_ticket_waves`
derives are the PRIMARY wave's price/currency and the SUM of every
wave's stock. The NIP-52 publisher read those, so as soon as an
organiser created a second wave the public card would advertise the
early-bird price after early bird closed and count stock in waves that
had not opened. Refs #61.

`build_nip52_event` now describes the wave a buyer can actually buy
from:

- several waves can be open at once, and a publisher has no one to ask
  which one the buyer wants (the purchase endpoint errors with "Please
  select a ticket wave"), so it advertises the CHEAPEST open wave — the
  price a buyer is able to obtain. Deviation recorded in
  docs/upstream-candidates.md.
- with no open wave, `tickets_available` is 0 and never omitted:
  omission used to mean "unlimited", which #34/#62 removed as a concept.
- `tickets_payment_methods` is scoped to the advertised wave too. It was
  derived from `event.allow_fiat` — the primary wave's — so it could
  offer a fiat rail while `tickets_allow_fiat` was absent and the
  purchase endpoint would refuse it. They are the same fact and now come
  from the same place.

Wave boundaries are time-driven, and every republish we have is
sale-driven, so nothing fires when early bird ends at midnight. Rather
than add a scheduler, a publish records which wave it advertised
(`nostr_published_wave_id`, m004) and the reconciliation sweep compares
that against the wave that would be advertised now, setting
`nostr_publish_pending` on a mismatch — reusing the existing retry path.
NULL means "never published", which the sweep leaves alone so an upgrade
does not republish the whole table on first boot.

The selection rule lives in `models.advertised_ticket_wave` so the
publisher and the drift detector cannot disagree about what is on the
relay.

17 new tests; 114 pass. ruff, black, prettier clean; mypy error set
still identical to HEAD's baseline.

docs/rebase-playbook.md

@@ -10,7 +10,7 @@ Modes C and D were added later in the same rebase, from `services.py`.
 
 ## The four failure modes
 
-Git resolves *text*. Neither of these produces a conflict marker.
+Git resolves _text_. Neither of these produces a conflict marker.
 
 ### A. Missed application
 
@@ -23,14 +23,14 @@ so the invariant silently does not hold there.
 > `get_event` and `get_events` in `crud.py`. Both merged cleanly. But the
 > fork has four getters upstream never had — `get_all_events`,
 > `get_public_events`, `get_pending_events`,
-> `get_events_pending_republish` — and two of them *publish*
+> `get_events_pending_republish` — and two of them _publish_
 > (`/republish-all` and the #55 sweep). Without the same call they emit
 > the stale roll-up, so waves would have been wrong on exactly the paths
 > that push to relays, and nowhere else.
 
 ### B. Changed meaning
 
-Upstream redefines what an existing field *means*. Fork code that reads
+Upstream redefines what an existing field _means_. Fork code that reads
 it is untouched by the diff and keeps compiling, while now saying
 something false.
 
@@ -44,7 +44,7 @@ something false.
 ### C. Misplaced conflict boundary
 
 Git anchors a conflict on whatever lines happen to match. When both sides
-rewrote the same region, an *incidental* shared line inside it can become
+rewrote the same region, an _incidental_ shared line inside it can become
 the anchor — and everything past that line lands **outside** the markers,
 where it reads as cleanly merged.
 
@@ -57,7 +57,7 @@ is upstream's — the fork's version is shadowed without a single warning.
 > Ours (220 lines: multipart HTML mail, the QR-card attachment, the
 > Date/Message-ID headers that keep SpamAssassin quiet, npub DM support)
 > appeared between the markers; upstream's showed as 4 lines. But both
-> sides define the *same nine functions*, and git had anchored on a
+> sides define the _same nine functions_, and git had anchored on a
 > shared `_send_nostr_ticket_notification` line — so upstream's entire
 > parallel stack sat below `>>>>>>>`, looking merged. Taking "ours" and
 > moving on would have left upstream's definitions last in the file and
@@ -102,7 +102,7 @@ histories.
 
 ## The procedure
 
-Run this *after* the merge resolves and *before* the release.
+Run this _after_ the merge resolves and _before_ the release.
 
 ### 1. Enumerate what upstream introduced
 
@@ -117,7 +117,7 @@ git show <tag>:models.py | sed -n '/^def sync_event_ticket_waves/,/return event/
 Split the result into **new symbols** (mode A candidates) and
 **redefined fields** (mode B candidates).
 
-### 2. For each new symbol upstream *calls*, find the fork-only siblings
+### 2. For each new symbol upstream _calls_, find the fork-only siblings
 
 Ask what category of place the call belongs to — "every function that
 returns an `Event` from the DB", "every path that prices a ticket" — then
@@ -147,7 +147,7 @@ grep -n "price_per_ticket\|amount_tickets" $(comm -23 ...)
 Both examples above were confirmed by reading the code path end to end,
 not inferred from the diff. A wrong theory costs more than the check:
 during this rebase an inference that `amount_tickets` "goes stale on
-sale" was wrong — `sync_event_ticket_waves` also runs on *reads*, which
+sale" was wrong — `sync_event_ticket_waves` also runs on _reads_, which
 only the call-site list showed.
 
 ### 5. Write the reason at the site