fix: don't double export aggregate efficiency metrics

[cgrinds]

This was discovered by log messages from Prometheus like this:

msg="Error on ingesting samples with different value but same timestamp"

after enabling debug logging in Prometheus more information is printed

Oct 24 04:24:07 harvest-ubuntu-03 prometheus[32020]: ts=2024-10-24T08:24:07.702Z caller=scrape.go:1859 level=debug component="scrape manager" scrape_pool=prometheus target=
http://localhost:13002/metrics
msg="Duplicate sample for timestamp" series="aggr_total_physical_used{aggr="umeng_aff300_aggr2",cluster="umeng-aff300-01-02",datacenter="nane",node="umeng-aff300-01"}"

This happened when you specified collectors in this order:

collectors:
  - Rest
  - Zapi

The Zapi collector exports aggregate efficiency objects via the AggregateEfficiency template while the REST collector exports them via the Aggregate template. That means the normal object/collector dedup logic won't handle these metrics and they are double exported.

This fix removes the double export and Prometheus ingest error.