Allow mechanism for breaking time series generation across multiple files (e.g. 10-year chunks)

[mnlevy1981]

Is your feature request related to a problem? Please describe.

The time series block of the key_metrics example looks like

timeseries:
  num_procs: 8
  ts_done: [False, False]
  overwrite_ts: [False, False]
  case_name: ['b.e30_beta02.BLT1850.ne30_t232.104', 'b.e23_alpha17f.BLT1850.ne30_t232.092']

  atm:
    vars: ['PSL']
    derive_vars: []
    hist_str: 'h0a'
    start_years: [1,1]
    end_years: [100,100]
    level: 'lev'

This will create time series for years 1-100 for both b.e30_beta02.BLT1850.ne30_t232.104 and b.e23_alpha17f.BLT1850.ne30_t232.092, but it would be really nice if we could create time series for years 1-10, 11-20, 21-30, ..., 91-100 instead.

Describe the solution you'd like
Support nested lists, such as

  atm:
    vars: ['PSL']
    derive_vars: []
    hist_str: 'h0a'
    start_years: [[1, 11, 21, 31, 41, 51, 61, 71, 81, 91], [1, 11, 21, 31, 41, 51, 61, 71, 81, 91]]
    end_years: [[10, 20, 30, 40, 50, 60, 70, 80, 90, 100], [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]]
    level: 'lev'

---

While I'm here, another improvement to the time series generation API that I'd like is dropping the requirement for

timeseries:
  case_name: ['b.e30_beta02.BLT1850.ne30_t232.104', 'b.e23_alpha17f.BLT1850.ne30_t232.092']

And letting case_name be determined from global_params instead, specifically the case_name and base_case_name variables:

global_params:
  case_name: 'b.e30_beta02.BLT1850.ne30_t232.104'
  base_case_name: 'b.e23_alpha17f.BLT1850.ne30_t232.092'

For this, we could do something like

if timeseries_case is None:
  timeseries_case = [global_params_case]
  if global_params_base_case is not None:
    timeseries_case.append(global_params_base_case)

then users could still specify timeseries.case_name in the YAML, but if it's omitted then it's determined from global_params instead.