Pick the per-day "Usage" CSVs and drop the "Audit" mirror. B2 names them
usage.account-<acct>.csv and usage.audit-account-<acct>.csv — BOTH contain
"usage", so the discriminator is the absence of "audit". Including the audit
file would double-count every metric. Falls back to all non-audit CSVs if a
different account type names its files differently (Groups/Locations rows lack
account_id/date and are dropped by mapRow anyway).
Parameters
keys: string[]
Candidate report object keys.
Returns string[]
The report object keys that should be treated as usage data.
Pick the per-day "Usage" CSVs and drop the "Audit" mirror. B2 names them
usage.account-<acct>.csvandusage.audit-account-<acct>.csv— BOTH contain "usage", so the discriminator is the absence of "audit". Including the audit file would double-count every metric. Falls back to all non-audit CSVs if a different account type names its files differently (Groups/Locations rows lack account_id/date and are dropped by mapRow anyway).