InfluxDB timestamp issues

[hagertnl]

There are two issues I wanted to bring up:

1. Do we really need nanosecond precision? InfluxDB docs claim that you should use the coarsest precision you can to get the best performance.
2. Currently, metrics & nodehealth entries do not set a timestamp when they log. If nanoseconds matter, they should set a timestamp. I think they should use the same timestamp as check_end event and log with a precision of milliseconds or coarser, at least for metrics & nodehealth entries, since there is only one entry per test_id.

[hagertnl]

Removed from v3 target fixes. This likely requires refactoring so that different precisions can be specified for each bucket (ie, nanosecond for events, millisecond for metrics and nodehealth).