[CDB-READER] Identify and review performance-critical operations

[StudioWEngineers]

Load and retrieve information of the same type for different load cases (e.g., cable results) from a cdb can be computationally expensive for two reasons:

1. large numbers of load cases to be loaded
2. resizing the DataFrame every time a load case is loaded
3. info are needed (most of the time) per single element and load case in a large number

The following improvements should be tested:

• define a parallel_load method which avoid manipulating the _data DateFrame at every load case
• use multiprocessing to load multiple load case at the same time
• use multi-indexing by ELEM_ID and LOAD_CASE