Standard weekly patch release

[1.3.2] - 2021-05-18

Changed

• DataModules now avoid duplicate {setup,teardown,prepare_data} calls for the same stage (#7238)

Fixed

• Fixed parsing of multiple training dataloaders (#7433)
• Fixed recursive passing of wrong_type keyword argument in pytorch_lightning.utilities.apply_to_collection (#7433)
• Fixed setting correct DistribType for ddp_cpu (spawn) backend (#7492)
• Fixed incorrect number of calls to LR scheduler when check_val_every_n_epoch > 1 (#7032)

Contributors

@alanhdu @carmocca @justusschock @tkng

If we forgot someone due to not matching commit email with GitHub account, let us know :]