Init the process group with a load-scaled timeout for sharded loading (#48228)
* Init the process group with a load-scaled timeout for sharded loading
Sharded loading of a TB-scale checkpoint takes tens of minutes and ranks
finish far apart (measured: >10 minutes spread on GLM-4.6, 714 GB over 16
ranks), so the fastest ranks' first collective dies on the default 10-minute
NCCL watchdog. Use a 2-hour timeout when transformers initializes the group
itself.
* Move the timeout comment onto the argument it explains