Add support for ingesting datasets having multiple dimensions in earth engine

[kbp45-cusp]

resolves #305

Solution approach:

As the dataset contains multiple dimensions apart from latitude and longitude, we need to convert the data arrays to 2D grids. Hence We need to remove the all other dimensions. So, we are going to accept a list of dimension from user, on which we will partition the data. There might be cases when user don't want to partition the data on the basis of all dimensions. In such cases, the remaining dimensions will be flattened out and final grid will be of 2D. Also, we need to allow user to specify the name of the resulting COGs based on the partitioned dimensions.

Key notes for each cli arguments:

• partition_dims: Dimensions list on basis of which the dataset needs to be partitioned. Let's say if a dataset having 5 dimensions (time, step, level, latitude, longitude) and user only provides [time, step] then flatten the 'level' dimension into variables.
• asset_name_format: This argument enables the user to decide the name of the resulting COG. The user will be providing a template string containing dimension names. Here, user needs to provide only the dimensions used to partition. User can add 'init_time' and 'valid_time' as dimension name, provided that dim_mapping is passed as well.
• Need for dim_mapping: This mostly handles the case when the 'step' dimension (or any dimension suggesting forecasting values) is not in datetime format (it is either in timedelta or an integer). So, this dictionary will contain mapping to 'init_time' and 'valid_time' dimension, so we will get to know to which dimension the timedelta needs to be added (basically which dimension serves as init_time). With these information, we can add the forecasting datetime in the asset_name as well (if user requires it) also we can add 'start_time' and 'end_time' values in the attributes which will then be ingested in earth engine.
• date_format: The format for the datetime values used in the asset_name.

[j9sh264]

Initial pass comments.