Partial eval optimizations leveraging enhanced static tracking

[idavis]

Once #2087 is resolved, we will be able to better track static vars through partial eval. This opens the door for further optimizations such as

• Identity ops for int (and double once Adaptive profile floating-point computations extension #2078 is merged)
  • 0 + var, var + 0
  • 1 + var, var + 1
  • var - 0
• Shortcuts
  • 0 * var, var * 0