Add more CSV coverage for time series aggregate functions

[limotova]

Right now most of these functions only have a few test cases in the csv tests (k8s-timeseries.csv-spec) but we'd like to have a few more test cases so we can have better backwards compatibility coverage, as well as just having some hardcoded test cases

Functions to be covered:

• rate
• avg_over_time
• count_over_time
• count_distinct_over_time
• min_over_time (Larisa)
• max_over_time (Larisa)
• sum_over_time
• first_over_time
• last_over_time

Examples of things to cover:

• All expected data types as input
• Grouping and non-grouping
• Known edge cases (*see non-OverTime tests for ideas as well)
• Any expected warnings (might not exist)
• Features expected to be used with the aggregation frequently (examples: where clauses within the stats command, eval on the results, etc)

[elasticsearchmachine]

Pinging @elastic/es-storage-engine (Team:StorageEngine)