We want our core StorageScheduler to handle losses over time, for better support of thermal storage.

• device_constraints in device_scheduler gets a new column (e.g. named efficiency in units of %) to describe how much stock is left after a time step. The scheduler sets up a corresponding parameter and uses it in device_bounds. Add a test for the new functionality.
• StorageScheduler gets a new argument (e.g. named storage_efficiency in units of %/<sensor resolution>) from which the new device_constraints column is derived (i.e. convertion of the storage efficiency to the event resolution of the scheduled sensor should be done by the client, using a compound interest calculation). *
• The new argument is exposed through the CLI.
• The new argument is exposed through the API, with user documentation.

* For example, a storage efficiency of 99 percent per (absolute) day, for scheduling a 1-hour resolution sensor, should be passed as a storage efficiency of $0.99^{1/24}$.