Allow non-linear rates of change between stages

[ArnoClaude]

Description

Support non-linear parameter evolution across stages instead of only constant growth rates.

Current Behavior

• User specifies scalar rate of change per parameter
• Parameters grow via compound rate: value_t = value_0 * (1 + rate)^t
• Example: demand=10000, rate=0.10 → stage 2: 11000, stage 3: 12100

Proposed Behavior

• User selects growth type from predefined options:
  • linear: value_0 + rate * t
  • compound: value_0 * (1 + rate)^t (current default)
  • exponential: value_0 * exp(rate * t)
  • custom: user provides list of explicit values per stage
• Alternatively: accept arbitrary expression string (e.g., "1.5 * x1 + 100")

Implementation Notes

• Predefined functions are safer/simpler than expression parsing
• custom list option handles any pattern without parsing complexity
• Config example:

  demand:
    base: 10000
    growth_type: compound  # or linear, exponential, custom
    rate: 0.10
    # OR for custom:
    # values: [10000, 12000, 15000, 20000]
  

Open Questions

• Predefined functions sufficient, or need arbitrary expressions?
• If arbitrary: use sympy for parsing? Security concerns?
• Should different parameters support different growth types in same run?