BUG/MINOR: contrib/prometheus-exporter: Restart labels dump at the right pos

BUG/MINOR: contrib/prometheus-exporter: Restart labels dump at the right pos

For some metrics, several lines are produced per entity, one per label
value. For instance, the health-check status (ST_F_CHECK_STATUS) or the
entity status (ST_F_STATUS). The dump may be stopped in the middle of the
labels processing if the output buffer is full. This means the next time, we
must take care to restart on the right label value.

For now, this part is buggy and we always restart to dump all the label
values again from the beginning. To be sure to restart at the right
position, the field <ctx.stats.st_code> in the applet context is used to
save the last position. Of course, we take care to reset this value when
necessary.

This fix is specific for 2.4. No backport needed.