Jetstream configs should support easy downsampling

[danielkberry]

Currently, if a job fails due to memory issues, a DS must manually override the enrollments query to do downsampling. This is complex, time-consuming, and error-prone. It would be helpful if support could be built for easy downsampling. Something like:

[experiments]
sample_id_range = [0,10]

to explicitly use 0 <= sample_id <= 10
or

[experiments]
sample_percentage = 75

to implicitly use sample_id <= 75

┆Issue is synchronized with this Jira Task

[mikewilli]

See related mozilla/mozanalysis#187

[mikewilli]

And mozilla/metric-config-parser#298

[mikewilli]

For posterity: the solution for this issue was to add the config option sample_size, which is an integer 1-100 representing the sample rate (e.g., sample_size = 20 would use a 20% sample).