Increase level of information and make consistent across plot file names and output displays

### What problem does your feature request solve?

With growing sophistication of CSET outputs, there would be value in improving the tracking of key information in plot filenames, and some plot outputs. For example, there is now a need to track some combinations of:
- time extent of inputs,
- domain of output (bearing in mind this could vary if SELECT_SUBAREA true),
- models in use
- variable shown,
- .....

Some but not all of these attributes are explicitly or implicitly evident visually in plot outputs (or legends), but some not. Relatively little of this information is then captured in plot output filenames (or equivalent data filenames). At present, traceability of plot inputs is managed via cset-workflow web directory structures (based on task names).

Therefore when running e.g. command-line, I get filenames like:
histogram_temperature_at_screen_level_479545.0.png
domain_mean_number_of_lightning_flashes_in_column_time_series_aggregation_by_hour_of_day.png
regardless of domain or time range of interest etc. This also risks outputs being overwritten if performing local analysis of many experiments etc.

### Describe the solution you'd like

Define a known, traceable, and preferably short naming convention that provides sufficient information to user of what is in file from filenames, and similarly ensure plots visually contain enough information to discern any key aspects of plot inputs.

It would make most sense to work this up after completion and merge of #1377

### Describe alternatives you've considered

Review the scope and outcome of #1377 first, then map where plot filenames and visual output metadata are produced. Define a short, consistent convention covering the relevant input and output attributes, and consider the work complete when filenames and plots make those attributes traceable without relying on workflow directory names.

Written by the indexing model from the issue text.