Improve Job Search IV: Add explanations, simulation, and cross-sectional analysis

Major improvements to Job Search IV (Fitted Value Function Iteration):

1. Notation and clarity:
   - Changed utility function from u(c, γ) to u(x, γ) to avoid confusion with
     unemployment compensation parameter c
   - Fixed random variable notation: w → W_t in model description
   - Changed z_i → Z_i in Monte Carlo formula
   - Updated Model constructor from named to positional arguments
   - Renamed variables in T and get_greedy: accept → v_e, reject → continuation_values
   - Standardized naming: h_star → h, w_star → w_bar

2. Added explanations:
   - Explained the explicit form of (Pv_u)(w) with derivation
   - Changed "From the equation above" to "Recall that we have"
   - Added explanation before Bellman operator about Monte Carlo integration

3. Updated plot styling:
   - Simplified plot to match Job Search III style
   - Show only continuation value h and employment value v_e
   - Updated labels and explanation text

4. Added Simulation section:
   - Implemented update_agent function with job separation logic
   - Added simulate_employment_path function
   - Created three-panel plot showing employment status, wage path, and
     cumulative unemployment rate
   - Properly handles Markov wage dynamics with separation

5. Added Cross-Sectional Analysis subsection:
   - Vectorized update function for parallel agent simulation
   - JIT-compiled simulation using lax.fori_loop
   - Cross-sectional unemployment histogram
   - Comparison of time-average vs cross-sectional unemployment rates

6. Added missing import: from functools import partial

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: jstac