[Stack Monitoring] Allow sub minute interval for CPU usage rule

[klacabane]

Summary

The cpu usage rule only supports minute as an interval unit and will convert any specified period to a minute. When dealing with a sub-minute interval this leads to a fraction and will break the execution with the following error:

x_content_parse_exception: [x_content_parse_exception] Reason: [1:852] [date_histogram] failed to parse field [fixed_interval], caused by: "failed to parse [0.9833333333333333m], fractional time values are not supported,For input string: "0.9833333333333333""

According to this comment it was a conscious decision but broke the seconds use case as a side effect.

We should investigate why converting the interval in ms with the date histogram aggregation doesn't work with this query and either fix the underlying issue or find an alternative that allows a sub-minute interval to be set.

AC

• The cpu usage rule works with intervals inferior to a minute

[elasticmachine]

Pinging @elastic/infra-monitoring-ui (Team:Infra Monitoring UI)

[miltonhultgren]

The error itself is just normal stuff from the date_histogram aggs, fractional units of any size are not supported.

Fractional time values are not supported, but you can address this by shifting to another time unit (e.g., 1.5h could instead be specified as 90m).

That comment is very vague but I wonder if it's due to rounding issues?
Beyond that I can't think of any reason why we couldn't translate 0.9m into 54s.