[FEA] Consider core-seconds in TCL heuristics

[cindyyuanjiang]

Is your feature request related to a problem? Please describe.
We found a job with low compute and ran on a cheaper instance type on Databricks does not run faster with GPU. We should factor in minimum core second threshold in Top Candidate View.

• [FEA] Add total core seconds in Qualification core tool output #1320
• [FEA] Add total core seconds into top candidate view #1342

[mattahrens]

Initial scope is to calculate core seconds and add it as a metric in summary output CSV file and possibly CLI output as well.

Calculation for core seconds is the total executor time (by finding start/end time for each executor to get each executor duration) multiplied by the spark.executor.cores value for the workload.

[cindyyuanjiang]

Closing as all subtasks are done.