feat(logging): add .With() system identifiers to all top-level loggers

[corylanou]

Summary

After fixing default slog usage in #1178, the next logical step is to add .With() context to all top-level loggers so log lines self-identify which system/subsystem produced them.

Current State

Some components already do this:

• DB: slog.With("db", filepath.Base(path))
• Replica: logger.With("replica", r.Client.Type())
• VFS: logger.With("vfs", "true") and .With("name", filename)

But Store, Server, Compactor, WALReader, and all replica clients log without system identification.

Proposed Hierarchy

Store:       slog.Default().With("system", "store")
  └─ DB:       store.Logger.With("db", "mydb.sqlite")           // already done
      └─ Replica:    db.Logger.With("replica", "s3")             // already done
          └─ Client:   replica.Logger().WithGroup("s3")          // needs parent propagation
      └─ Compactor:  db.Logger.With("subsystem", "compactor")   // needs adding
      └─ WALReader:  db.Logger.With("subsystem", "wal-reader")  // needs adding

Server:      slog.Default().With("system", "server")

Example log output:

level=INFO msg="compacted snapshots" system=store db=mydb.sqlite subsystem=compactor

Key Changes Needed

Tier 1 — Top-level (Store, Server)

• store.go: s.Logger = slog.Default().With("system", "store")
• server.go: logger = slog.Default().With("system", "server")

Tier 2 — Subsystems (Compactor, WALReader)

• Add .With("subsystem", "compactor") and .With("subsystem", "wal-reader")

Tier 3 — Replica clients (structural change)

• All 8 replica clients (s3, file, sftp, gs, webdav, oss, abs, nats) create loggers from slog.Default() instead of inheriting from their parent Replica/DB
• Need to accept a parent logger parameter (or SetLogger method) so context flows through the hierarchy
• S3 Leaser (s3/leaser.go) has the same issue

Benefits

• Every log line identifies its source system without manual inspection
• Context accumulates through the hierarchy (a compactor log includes the DB name)
• Easier debugging in multi-database setups