Today, the validation binary returns duplicate key errors as an exhaustive list of all affected documents. While this is handy, it can exceed size limits for downstream users of the command.

Unique index 'a-1' has duplicate key: { : a, : 1}, first record: RecordId(1), second record: RecordId(2)"

We should consider a more concise error that simply says:

"Unique index .* has n duplicate keys.  See log id xxx for more information"