🚀 The feature, motivation and pitch

Context of the issues are in #90212

Current State: The init() method does the setup work for both SavePlanner and LoadPlanner. For example, this determines how the state_dict for distributed checkpoint should be set up (see example here https://github.com/pytorch/pytorch/blob/master/torch/distributed/checkpoint/planner.py#L126-L133). We also have a init() method in DefaultSavePlanner and DefaultLoadPlanner to control a few flags (flatten_nested_tensor and flatten_state_dict). The two similar names could be confusing for users. We should rename init() method to something else to avoid the confusion.

Change proposed: Rename init() method across the board.

Alternatives

No response

Additional context

No response

cc @mrshenli @pritamdamania87 @zhaojuanmao @satgera @rohan-varma @gqchen @aazzolini @osalpekar @jiayisuse @H-Huang @kwen2501 @awgu