Add _txlock=immediate so batch writes wait instead of dropping (closes #25)

Batch flush paths read before they write, so a deferred transaction starts
as a reader and must upgrade to the write lock on its first INSERT/UPDATE.
When the background maintainer holds the write lock for a WAL checkpoint,
that upgrade fails immediately with "database is locked" and the busy
timeout does not apply, so the batch is dropped. Adding _txlock=immediate
to the DSN makes every transaction take the write lock at BEGIN, so it
waits up to busy_timeout instead of failing.

A regression test drives batch writes against a running checkpoint loop and
fails with "database is locked" without the change.

Model: opus-4-8

internal/database/database_test.go

@@ -4,7 +4,9 @@ import (
 	"context"
 	"database/sql"
 	"net"
+	"sync"
 	"testing"
+	"time"
 
 	"git.eeqj.de/sneak/routewatch/internal/config"
 	"git.eeqj.de/sneak/routewatch/internal/logger"
@@ -18,6 +20,18 @@ const tempStoreMemory = 2
 // once so each is a distinct SQLite connection that parsed the DSN.
 const heldConnections = 5
 
+// Parameters for the checkpoint-contention regression test.
+const (
+	// contentionIterations is how many batch writes race the checkpoint loop.
+	contentionIterations = 400
+	// contendedASNCount is the small set of ASNs the batches reuse, so most
+	// batches update existing rows and exercise the read-before-write path.
+	contendedASNCount = 16
+	// asnSecondBand offsets a second ASN per batch so each batch writes more
+	// than one row.
+	asnSecondBand = 100
+)
+
 func TestIPToUint32(t *testing.T) {
 	tests := []struct {
 		name     string
@@ -361,6 +375,56 @@ func TestConnectionPoolPragmas(t *testing.T) {
 	}
 }
 
+// TestBatchWriteDuringCheckpoint reproduces issue #25. A batch write reads
+// (SELECT) before it writes (INSERT/UPDATE). Under the default deferred locking
+// the transaction begins as a reader and, when the maintainer's WAL checkpoint
+// holds the write lock, its upgrade to writer fails immediately with "database
+// is locked" without honouring busy_timeout, dropping the batch. With
+// _txlock=immediate the transaction takes the write lock at BEGIN and waits, so
+// no batch is dropped. The checkpoint runs without the Database mutex, exactly
+// as the background maintainer does in production.
+func TestBatchWriteDuringCheckpoint(t *testing.T) {
+	cfg := &config.Config{StateDir: t.TempDir()}
+
+	db, err := New(cfg, logger.New())
+	if err != nil {
+		t.Fatalf("failed to create database: %v", err)
+	}
+	defer func() { _ = db.Close() }()
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	var wg sync.WaitGroup
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			default:
+				_ = db.Checkpoint(ctx) // errors are the checkpoint's own to absorb
+			}
+		}
+	}()
+
+	ts := time.Now().UTC()
+	for i := 0; i < contentionIterations; i++ {
+		asns := map[int]time.Time{
+			i % contendedASNCount:                   ts,
+			(i % contendedASNCount) + asnSecondBand: ts,
+		}
+		if err := db.GetOrCreateASNBatch(asns); err != nil {
+			cancel()
+			wg.Wait()
+			t.Fatalf("batch write failed under checkpoint contention: %v", err)
+		}
+	}
+
+	cancel()
+	wg.Wait()
+}
+
 func BenchmarkIPToUint32(b *testing.B) {
 	ip := net.ParseIP("192.168.1.1")
 	b.ResetTimer()