tools/cephfs_mirror: Fix failed-to-syncing state priority across metrics paths

When a mirrored directory hits the consecutive failure threshold, the
failed flag stays set until the next successful sync. On retry, the
directory can be actively syncing while failed remains true, but
dump_sync_stat() and update_directory_current_sync_perf_counters()
checked failed before current_syncing_snap and kept reporting "failed".

Introduce get_dir_sync_state() and use it in peer_status, omap
persistence, and per-directory perf counters so syncing takes precedence
over the sticky failed flag.

Fixes: https://tracker.ceph.com/issues/76616
Signed-off-by: Kotresh HR <khiremat@redhat.com>