nans should be handled by hydraulic fields due to use of DataFrames

User Story: As mentioned in issue an error is returned saying that there are nans in the data. During the model initialization, the hydraulic sample data is transformed between json and DataFrames. As data fields are not populated all the time, sometimes columns have 'nan' values.

Acceptance Criteria: The model worker should run with data (specifically data from section 12 on 16/3/2022 at about 1pm) where the population level of the data changes. Proposed is to alter the datamodel to allow 'nan' in the schema, and then delete any fields assigned this value, treating it as if unpopulated.