Consider a "partial" availability analysis for datasets without all the necessary inputs

Is your feature request related to a problem? Please describe.
The AvailabilityAnalysis class added in #131 requires four inputs: system power, subsystem power, system cumulative energy, and system expected power. The AvailabilityAnalysis class implements two separate analysis strategies, each of which needs only a subset of the three measured fields, but there is no way to only run one of the strategies for datasets that don't have all the fields.

Describe the solution you'd like
Some way to provide the class with only a subset of the inputs and run whichever strategy uses only those inputs. E.g. if a system doesn't have cumulative energy data, then run whatever analysis can be done without it instead of failing immediately.

Describe alternatives you've considered
None

Additional context
The notebook currently shows one way of "getting around" this requirement, but it's sort of hacky and is only half of the desired solution -- it only gives results for loss_system, but the other way round (only giving results for loss_subsystem) isn't possible.

Start with the AvailabilityAnalysis class introduced in #131 and the system_availability_example notebook's “Other use-cases” section. Trace which inputs each analysis strategy requires and how missing fields currently cause failure. Done means datasets with either supported subset can run the applicable strategy and produce the corresponding loss result without the notebook workaround.