[SWDEV-555718] - Remerge backport of inductor autotune configs (#2668)

Cherry picks of inductor autotune caused some breakages, gone through
this process again ensuring other required commits are picked.

---------

Co-authored-by: Bert Maher <[email protected]>
Co-authored-by: PaulZhang12 <[email protected]>
Co-authored-by: Nikhil Anil Patel <[email protected]>

Author: 4 people

torch/_inductor/config.py

@@ -410,9 +410,6 @@ def prologue_fusion_enabled() -> bool:
     os.environ.get("TORCHINDUCTOR_AUTOTUNE_FALLBACK_TO_ATEN", "1") == "1"
 )
 
-# DEPRECATED. This setting is ignored.
-autotune_fallback_to_aten = False
-
 # the value used as a fallback for the unbacked SymInts
 # that can appear in the input shapes (e.g., in autotuning)
 unbacked_symint_fallback = 8192