[xegpu] Elementwise schedule and refactoring

[tkarna]

• Adds a simple linalg-to-binary schedule for elementwise kernels.
• Refactors common parts with the mlp schedule to sub-schedules (python functions in lowering_common.py).
  • These sub-schedules could be outlined into separate schedules if needed.
• The mlp and elementwise schedules now take the payload function name for more robust matching.
  • GPU specific transformations should not be applied to the entire module nor all functions.