srrdb: harvest the srrDB search API into a separate SQLite database

New scope reading the API instead of scraping HTML: one search path per run,
passed raw through --single-uri, and the answer stored as-is into its own
SQLite file created from the embedded schema.

Command line: --single-uri carries the path after /v1/search/
(category:flac/date:today/page:1) and must name a category; --sleep sets the
pause between pages, with the config value as fallback. A day named on the
command line is always fetched again and its dump rewritten, so cron can
refresh today hourly and yesterday daily.

Date placeholders: date:auto walks backwards to the newest day in
[srrdb.floor..yesterday] that has no dump yet, closing holes instead of
declaring the archive covered. date:today, date:yesterday and date:tomorrow
stand for the neighbouring days. Today is never taken by the walk, because a
day still being filled would be frozen by its own dump after the first run.

Dumps are the cursor as well as the record, so a missing dump directory is an
error instead of "nothing harvested yet": a renamed, unmounted or mistyped
directory must not send the run silently back to the beginning. An empty day is
a normal result and is remembered as harvested; API warnings never are, and
they reach the exit code.

Writes go through one transaction, so a day lands completely or not at all.
group and category come from the release line and the search path; status, nfo,
content and synced_at stay NULL. srrdb.log.jsonl records one line per finished
run, srrdb.error_log.jsonl keeps failures and API warnings apart by a kind
field.

Also: the console command contract returns (string, error), so a failing scope
reaches the process exit code and cron can tell a failed run from a successful
one. config/scope/srrdb.yaml is untracked because its paths differ per host.
modernc.org/sqlite replaces the archived libsql client, which only ever talked
to a server. pkg/repository/srrdb.go is split by responsibility and covered by
table-driven tests.

pkg/repository/srrdb.go

@@ -0,0 +1,140 @@
+package repository
+
+import (
+	"errors"
+	"fmt"
+	"log/slog"
+	"strings"
+	"time"
+
+	"github.com/jmoiron/sqlx"
+
+	"git.kplus.net.ua/yevhen/resource-scraper/pkg/repository/table"
+	"git.kplus.net.ua/yevhen/resource-scraper/types/model"
+)
+
+// srrdbTimestamp keeps created_at in the same "YYYY-MM-DD HH:MM:SS" layout the
+// API itself uses in its date field, because the column is TEXT.
+const srrdbTimestamp = "2006-01-02 15:04:05"
+
+// Srrdb harvests the srrDB search API into the separate SQLite database.
+type Srrdb struct {
+	db *sqlx.DB
+}
+
+// NewSrrdbRepository returns a collector writing into the SQLite handle. The
+// handle is nil for every scope except srrdb, because Bootstrap connects the
+// SQLite database lazily, so the collector refuses to run without it instead of
+// failing later on a nil handle.
+func NewSrrdbRepository(db *sqlx.DB) *Srrdb {
+	return &Srrdb{db: db}
+}
+
+// Search harvests one search path. Nothing is written to the database until the
+// whole set is collected and its size matches what the API reported, so the
+// dump and the success log can only ever describe a complete day.
+func (s *Srrdb) Search(rawURI string) (int, error) {
+	if s.db == nil {
+		err := errors.New("srrdb collector has no database handle")
+		srrdbLogError(rawURI, err)
+
+		return 0, err
+	}
+
+	if !strings.Contains(rawURI, "category:") {
+		err := fmt.Errorf("search path %q has no category: segment", rawURI)
+		srrdbLogError(rawURI, err)
+
+		return 0, err
+	}
+
+	auto := strings.Contains(rawURI, srrdbAutoSegment)
+
+	now := time.Now()
+
+	uri, err := srrdbResolveAutoDate(srrdbResolveRelativeDays(rawURI, now), now)
+
+	switch {
+	case errors.Is(err, errSrrdbFloor):
+		slog.Info("srrdb backfill complete, srrdb.floor reached", "uri", rawURI)
+
+		return 0, nil
+	case err != nil:
+		srrdbLogError(rawURI, err)
+
+		return 0, err
+	}
+
+	// A slash typed on either end of the command line would build a double
+	// separator in the URL and an odd dump file name.
+	uri = strings.Trim(uri, "/")
+
+	// The guard applies to the automatic walk only. A day named on the command
+	// line is fetched again and its dump rewritten every time: a day can still
+	// gain releases later, either from another time zone or from a late upload,
+	// and cron refreshes today and yesterday on purpose.
+	dump := srrdbDumpPath(uri)
+
+	if auto {
+		exists, err := srrdbExists(dump)
+		if err != nil {
+			srrdbLogError(uri, err)
+
+			return 0, err
+		}
+
+		if exists {
+			slog.Info("srrdb already harvested, skipping", "uri", uri, "dump", dump)
+
+			return 0, nil
+		}
+	}
+
+	collected, err := srrdbCollectAll(uri)
+	if err != nil {
+		srrdbLogError(uri, err)
+
+		return 0, err
+	}
+
+	// An empty day is a normal outcome: zero results and no warnings. The dump
+	// records it as harvested, which keeps the walk moving. Warnings mean the
+	// path was not understood, and that must never be remembered as done.
+	switch {
+	case len(collected.Warnings) > 0:
+		srrdbLogWarning(uri, collected.Warnings)
+
+		return 0, fmt.Errorf("srrdb %q reported warnings: %s",
+			uri, strings.Join(collected.Warnings, "; "))
+	case len(collected.Results) != collected.ResultsCount:
+		err := fmt.Errorf("srrdb %q pulled %d of %d reported releases",
+			uri, len(collected.Results), collected.ResultsCount)
+		srrdbLogError(uri, err)
+
+		return 0, err
+	}
+
+	run := time.Now().Format(srrdbTimestamp)
+	category := srrdbCategory(uri)
+	rows := make([]model.SrrdbRelease, 0, len(collected.Results))
+
+	for _, result := range collected.Results {
+		rows = append(rows, srrdbRow(result, run, category))
+	}
+
+	if err := table.UpsertAllSrrdb(s.db, rows); err != nil {
+		srrdbLogError(uri, err)
+
+		return 0, err
+	}
+
+	if err := srrdbWriteDump(dump, collected); err != nil {
+		srrdbLogError(uri, err)
+
+		return len(rows), err
+	}
+
+	srrdbLogStatus(uri, len(rows), collected.ResultsCount)
+
+	return len(rows), nil
+}